~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to bzrlib/workingtree_4.py

  • Committer: Canonical.com Patch Queue Manager
  • Date: 2008-07-17 08:29:40 UTC
  • mfrom: (3549.1.4 stacking)
  • Revision ID: pqm@pqm.ubuntu.com-20080717082940-zdwz5cqhdoot1swx
(mbp) stacking post-merge review tweaks

Show diffs side-by-side

added added

removed removed

Lines of Context:
 
1
# Copyright (C) 2005, 2006, 2007, 2008 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
        basedir = safe_unicode(basedir)
 
135
        mutter("opening working tree %r", basedir)
 
136
        self._branch = branch
 
137
        self.basedir = realpath(basedir)
 
138
        # if branch is at our basedir and is a format 6 or less
 
139
        # assume all other formats have their own control files.
 
140
        self._control_files = _control_files
 
141
        self._transport = self._control_files._transport
 
142
        self._dirty = None
 
143
        #-------------
 
144
        # during a read or write lock these objects are set, and are
 
145
        # None the rest of the time.
 
146
        self._dirstate = None
 
147
        self._inventory = None
 
148
        #-------------
 
149
        self._setup_directory_is_tree_reference()
 
150
        self._detect_case_handling()
 
151
        self._rules_searcher = None
 
152
 
 
153
    @needs_tree_write_lock
 
154
    def _add(self, files, ids, kinds):
 
155
        """See MutableTree._add."""
 
156
        state = self.current_dirstate()
 
157
        for f, file_id, kind in zip(files, ids, kinds):
 
158
            f = f.strip('/')
 
159
            if self.path2id(f):
 
160
                # special case tree root handling.
 
161
                if f == '' and self.path2id(f) == ROOT_ID:
 
162
                    state.set_path_id('', generate_ids.gen_file_id(f))
 
163
                continue
 
164
            if file_id is None:
 
165
                file_id = generate_ids.gen_file_id(f)
 
166
            # deliberately add the file with no cached stat or sha1
 
167
            # - on the first access it will be gathered, and we can
 
168
            # always change this once tests are all passing.
 
169
            state.add(f, file_id, kind, None, '')
 
170
        self._make_dirty(reset_inventory=True)
 
171
 
 
172
    def _make_dirty(self, reset_inventory):
 
173
        """Make the tree state dirty.
 
174
 
 
175
        :param reset_inventory: True if the cached inventory should be removed
 
176
            (presuming there is one).
 
177
        """
 
178
        self._dirty = True
 
179
        if reset_inventory and self._inventory is not None:
 
180
            self._inventory = None
 
181
 
 
182
    @needs_tree_write_lock
 
183
    def add_reference(self, sub_tree):
 
184
        # use standard implementation, which calls back to self._add
 
185
        # 
 
186
        # So we don't store the reference_revision in the working dirstate,
 
187
        # it's just recorded at the moment of commit. 
 
188
        self._add_reference(sub_tree)
 
189
 
 
190
    def break_lock(self):
 
191
        """Break a lock if one is present from another instance.
 
192
 
 
193
        Uses the ui factory to ask for confirmation if the lock may be from
 
194
        an active process.
 
195
 
 
196
        This will probe the repository for its lock as well.
 
197
        """
 
198
        # if the dirstate is locked by an active process, reject the break lock
 
199
        # call.
 
200
        try:
 
201
            if self._dirstate is None:
 
202
                clear = True
 
203
            else:
 
204
                clear = False
 
205
            state = self._current_dirstate()
 
206
            if state._lock_token is not None:
 
207
                # we already have it locked. sheese, cant break our own lock.
 
208
                raise errors.LockActive(self.basedir)
 
209
            else:
 
210
                try:
 
211
                    # try for a write lock - need permission to get one anyhow
 
212
                    # to break locks.
 
213
                    state.lock_write()
 
214
                except errors.LockContention:
 
215
                    # oslocks fail when a process is still live: fail.
 
216
                    # TODO: get the locked lockdir info and give to the user to
 
217
                    # assist in debugging.
 
218
                    raise errors.LockActive(self.basedir)
 
219
                else:
 
220
                    state.unlock()
 
221
        finally:
 
222
            if clear:
 
223
                self._dirstate = None
 
224
        self._control_files.break_lock()
 
225
        self.branch.break_lock()
 
226
 
 
227
    def _comparison_data(self, entry, path):
 
228
        kind, executable, stat_value = \
 
229
            WorkingTree3._comparison_data(self, entry, path)
 
230
        # it looks like a plain directory, but it's really a reference -- see
 
231
        # also kind()
 
232
        if (self._repo_supports_tree_reference and
 
233
            kind == 'directory' and
 
234
            self._directory_is_tree_reference(path)):
 
235
            kind = 'tree-reference'
 
236
        return kind, executable, stat_value
 
237
 
 
238
    @needs_write_lock
 
239
    def commit(self, message=None, revprops=None, *args, **kwargs):
 
240
        # mark the tree as dirty post commit - commit
 
241
        # can change the current versioned list by doing deletes.
 
242
        result = WorkingTree3.commit(self, message, revprops, *args, **kwargs)
 
243
        self._make_dirty(reset_inventory=True)
 
244
        return result
 
245
 
 
246
    def current_dirstate(self):
 
247
        """Return the current dirstate object.
 
248
 
 
249
        This is not part of the tree interface and only exposed for ease of
 
250
        testing.
 
251
 
 
252
        :raises errors.NotWriteLocked: when not in a lock.
 
253
        """
 
254
        self._must_be_locked()
 
255
        return self._current_dirstate()
 
256
 
 
257
    def _current_dirstate(self):
 
258
        """Internal function that does not check lock status.
 
259
 
 
260
        This is needed for break_lock which also needs the dirstate.
 
261
        """
 
262
        if self._dirstate is not None:
 
263
            return self._dirstate
 
264
        local_path = self.bzrdir.get_workingtree_transport(None
 
265
            ).local_abspath('dirstate')
 
266
        self._dirstate = dirstate.DirState.on_file(local_path)
 
267
        return self._dirstate
 
268
 
 
269
    def filter_unversioned_files(self, paths):
 
270
        """Filter out paths that are versioned.
 
271
 
 
272
        :return: set of paths.
 
273
        """
 
274
        # TODO: make a generic multi-bisect routine roughly that should list
 
275
        # the paths, then process one half at a time recursively, and feed the
 
276
        # results of each bisect in further still
 
277
        paths = sorted(paths)
 
278
        result = set()
 
279
        state = self.current_dirstate()
 
280
        # TODO we want a paths_to_dirblocks helper I think
 
281
        for path in paths:
 
282
            dirname, basename = os.path.split(path.encode('utf8'))
 
283
            _, _, _, path_is_versioned = state._get_block_entry_index(
 
284
                dirname, basename, 0)
 
285
            if not path_is_versioned:
 
286
                result.add(path)
 
287
        return result
 
288
 
 
289
    def flush(self):
 
290
        """Write all cached data to disk."""
 
291
        if self._control_files._lock_mode != 'w':
 
292
            raise errors.NotWriteLocked(self)
 
293
        self.current_dirstate().save()
 
294
        self._inventory = None
 
295
        self._dirty = False
 
296
 
 
297
    @needs_tree_write_lock
 
298
    def _gather_kinds(self, files, kinds):
 
299
        """See MutableTree._gather_kinds."""
 
300
        for pos, f in enumerate(files):
 
301
            if kinds[pos] is None:
 
302
                kinds[pos] = self._kind(f)
 
303
 
 
304
    def _generate_inventory(self):
 
305
        """Create and set self.inventory from the dirstate object.
 
306
        
 
307
        This is relatively expensive: we have to walk the entire dirstate.
 
308
        Ideally we would not, and can deprecate this function.
 
309
        """
 
310
        #: uncomment to trap on inventory requests.
 
311
        # import pdb;pdb.set_trace()
 
312
        state = self.current_dirstate()
 
313
        state._read_dirblocks_if_needed()
 
314
        root_key, current_entry = self._get_entry(path='')
 
315
        current_id = root_key[2]
 
316
        if not (current_entry[0][0] == 'd'): # directory
 
317
            raise AssertionError(current_entry)
 
318
        inv = Inventory(root_id=current_id)
 
319
        # Turn some things into local variables
 
320
        minikind_to_kind = dirstate.DirState._minikind_to_kind
 
321
        factory = entry_factory
 
322
        utf8_decode = cache_utf8._utf8_decode
 
323
        inv_byid = inv._byid
 
324
        # we could do this straight out of the dirstate; it might be fast
 
325
        # and should be profiled - RBC 20070216
 
326
        parent_ies = {'' : inv.root}
 
327
        for block in state._dirblocks[1:]: # skip the root
 
328
            dirname = block[0]
 
329
            try:
 
330
                parent_ie = parent_ies[dirname]
 
331
            except KeyError:
 
332
                # all the paths in this block are not versioned in this tree
 
333
                continue
 
334
            for key, entry in block[1]:
 
335
                minikind, link_or_sha1, size, executable, stat = entry[0]
 
336
                if minikind in ('a', 'r'): # absent, relocated
 
337
                    # a parent tree only entry
 
338
                    continue
 
339
                name = key[1]
 
340
                name_unicode = utf8_decode(name)[0]
 
341
                file_id = key[2]
 
342
                kind = minikind_to_kind[minikind]
 
343
                inv_entry = factory[kind](file_id, name_unicode,
 
344
                                          parent_ie.file_id)
 
345
                if kind == 'file':
 
346
                    # This is only needed on win32, where this is the only way
 
347
                    # we know the executable bit.
 
348
                    inv_entry.executable = executable
 
349
                    # not strictly needed: working tree
 
350
                    #inv_entry.text_size = size
 
351
                    #inv_entry.text_sha1 = sha1
 
352
                elif kind == 'directory':
 
353
                    # add this entry to the parent map.
 
354
                    parent_ies[(dirname + '/' + name).strip('/')] = inv_entry
 
355
                elif kind == 'tree-reference':
 
356
                    if not self._repo_supports_tree_reference:
 
357
                        raise AssertionError(
 
358
                            "repository of %r "
 
359
                            "doesn't support tree references "
 
360
                            "required by entry %r"
 
361
                            % (self, name))
 
362
                    inv_entry.reference_revision = link_or_sha1 or None
 
363
                elif kind != 'symlink':
 
364
                    raise AssertionError("unknown kind %r" % kind)
 
365
                # These checks cost us around 40ms on a 55k entry tree
 
366
                if file_id in inv_byid:
 
367
                    raise AssertionError('file_id %s already in'
 
368
                        ' inventory as %s' % (file_id, inv_byid[file_id]))
 
369
                if name_unicode in parent_ie.children:
 
370
                    raise AssertionError('name %r already in parent'
 
371
                        % (name_unicode,))
 
372
                inv_byid[file_id] = inv_entry
 
373
                parent_ie.children[name_unicode] = inv_entry
 
374
        self._inventory = inv
 
375
 
 
376
    def _get_entry(self, file_id=None, path=None):
 
377
        """Get the dirstate row for file_id or path.
 
378
 
 
379
        If either file_id or path is supplied, it is used as the key to lookup.
 
380
        If both are supplied, the fastest lookup is used, and an error is
 
381
        raised if they do not both point at the same row.
 
382
        
 
383
        :param file_id: An optional unicode file_id to be looked up.
 
384
        :param path: An optional unicode path to be looked up.
 
385
        :return: The dirstate row tuple for path/file_id, or (None, None)
 
386
        """
 
387
        if file_id is None and path is None:
 
388
            raise errors.BzrError('must supply file_id or path')
 
389
        state = self.current_dirstate()
 
390
        if path is not None:
 
391
            path = path.encode('utf8')
 
392
        return state._get_entry(0, fileid_utf8=file_id, path_utf8=path)
 
393
 
 
394
    def get_file_sha1(self, file_id, path=None, stat_value=None):
 
395
        # check file id is valid unconditionally.
 
396
        entry = self._get_entry(file_id=file_id, path=path)
 
397
        if entry[0] is None:
 
398
            raise errors.NoSuchId(self, file_id)
 
399
        if path is None:
 
400
            path = pathjoin(entry[0][0], entry[0][1]).decode('utf8')
 
401
 
 
402
        file_abspath = self.abspath(path)
 
403
        state = self.current_dirstate()
 
404
        if stat_value is None:
 
405
            try:
 
406
                stat_value = os.lstat(file_abspath)
 
407
            except OSError, e:
 
408
                if e.errno == errno.ENOENT:
 
409
                    return None
 
410
                else:
 
411
                    raise
 
412
        link_or_sha1 = state.update_entry(entry, file_abspath,
 
413
                                          stat_value=stat_value)
 
414
        if entry[1][0][0] == 'f':
 
415
            return link_or_sha1
 
416
        return None
 
417
 
 
418
    def _get_inventory(self):
 
419
        """Get the inventory for the tree. This is only valid within a lock."""
 
420
        if 'evil' in debug.debug_flags:
 
421
            trace.mutter_callsite(2,
 
422
                "accessing .inventory forces a size of tree translation.")
 
423
        if self._inventory is not None:
 
424
            return self._inventory
 
425
        self._must_be_locked()
 
426
        self._generate_inventory()
 
427
        return self._inventory
 
428
 
 
429
    inventory = property(_get_inventory,
 
430
                         doc="Inventory of this Tree")
 
431
 
 
432
    @needs_read_lock
 
433
    def get_parent_ids(self):
 
434
        """See Tree.get_parent_ids.
 
435
        
 
436
        This implementation requests the ids list from the dirstate file.
 
437
        """
 
438
        return self.current_dirstate().get_parent_ids()
 
439
 
 
440
    def get_reference_revision(self, file_id, path=None):
 
441
        # referenced tree's revision is whatever's currently there
 
442
        return self.get_nested_tree(file_id, path).last_revision()
 
443
 
 
444
    def get_nested_tree(self, file_id, path=None):
 
445
        if path is None:
 
446
            path = self.id2path(file_id)
 
447
        # else: check file_id is at path?
 
448
        return WorkingTree.open(self.abspath(path))
 
449
 
 
450
    @needs_read_lock
 
451
    def get_root_id(self):
 
452
        """Return the id of this trees root"""
 
453
        return self._get_entry(path='')[0][2]
 
454
 
 
455
    def has_id(self, file_id):
 
456
        state = self.current_dirstate()
 
457
        row, parents = self._get_entry(file_id=file_id)
 
458
        if row is None:
 
459
            return False
 
460
        return osutils.lexists(pathjoin(
 
461
                    self.basedir, row[0].decode('utf8'), row[1].decode('utf8')))
 
462
 
 
463
    @needs_read_lock
 
464
    def id2path(self, file_id):
 
465
        "Convert a file-id to a path."
 
466
        state = self.current_dirstate()
 
467
        entry = self._get_entry(file_id=file_id)
 
468
        if entry == (None, None):
 
469
            raise errors.NoSuchId(tree=self, file_id=file_id)
 
470
        path_utf8 = osutils.pathjoin(entry[0][0], entry[0][1])
 
471
        return path_utf8.decode('utf8')
 
472
 
 
473
    def _is_executable_from_path_and_stat_from_basis(self, path, stat_result):
 
474
        entry = self._get_entry(path=path)
 
475
        if entry == (None, None):
 
476
            return False # Missing entries are not executable
 
477
        return entry[1][0][3] # Executable?
 
478
 
 
479
    if not osutils.supports_executable():
 
480
        def is_executable(self, file_id, path=None):
 
481
            """Test if a file is executable or not.
 
482
 
 
483
            Note: The caller is expected to take a read-lock before calling this.
 
484
            """
 
485
            entry = self._get_entry(file_id=file_id, path=path)
 
486
            if entry == (None, None):
 
487
                return False
 
488
            return entry[1][0][3]
 
489
 
 
490
        _is_executable_from_path_and_stat = \
 
491
            _is_executable_from_path_and_stat_from_basis
 
492
    else:
 
493
        def is_executable(self, file_id, path=None):
 
494
            """Test if a file is executable or not.
 
495
 
 
496
            Note: The caller is expected to take a read-lock before calling this.
 
497
            """
 
498
            self._must_be_locked()
 
499
            if not path:
 
500
                path = self.id2path(file_id)
 
501
            mode = os.lstat(self.abspath(path)).st_mode
 
502
            return bool(stat.S_ISREG(mode) and stat.S_IEXEC & mode)
 
503
 
 
504
    def all_file_ids(self):
 
505
        """See Tree.iter_all_file_ids"""
 
506
        self._must_be_locked()
 
507
        result = set()
 
508
        for key, tree_details in self.current_dirstate()._iter_entries():
 
509
            if tree_details[0][0] in ('a', 'r'): # relocated
 
510
                continue
 
511
            result.add(key[2])
 
512
        return result
 
513
 
 
514
    @needs_read_lock
 
515
    def __iter__(self):
 
516
        """Iterate through file_ids for this tree.
 
517
 
 
518
        file_ids are in a WorkingTree if they are in the working inventory
 
519
        and the working file exists.
 
520
        """
 
521
        result = []
 
522
        for key, tree_details in self.current_dirstate()._iter_entries():
 
523
            if tree_details[0][0] in ('a', 'r'): # absent, relocated
 
524
                # not relevant to the working tree
 
525
                continue
 
526
            path = pathjoin(self.basedir, key[0].decode('utf8'), key[1].decode('utf8'))
 
527
            if osutils.lexists(path):
 
528
                result.append(key[2])
 
529
        return iter(result)
 
530
 
 
531
    def iter_references(self):
 
532
        if not self._repo_supports_tree_reference:
 
533
            # When the repo doesn't support references, we will have nothing to
 
534
            # return
 
535
            return
 
536
        for key, tree_details in self.current_dirstate()._iter_entries():
 
537
            if tree_details[0][0] in ('a', 'r'): # absent, relocated
 
538
                # not relevant to the working tree
 
539
                continue
 
540
            if not key[1]:
 
541
                # the root is not a reference.
 
542
                continue
 
543
            relpath = pathjoin(key[0].decode('utf8'), key[1].decode('utf8'))
 
544
            try:
 
545
                if self._kind(relpath) == 'tree-reference':
 
546
                    yield relpath, key[2]
 
547
            except errors.NoSuchFile:
 
548
                # path is missing on disk.
 
549
                continue
 
550
 
 
551
    def kind(self, file_id):
 
552
        """Return the kind of a file.
 
553
 
 
554
        This is always the actual kind that's on disk, regardless of what it
 
555
        was added as.
 
556
 
 
557
        Note: The caller is expected to take a read-lock before calling this.
 
558
        """
 
559
        relpath = self.id2path(file_id)
 
560
        if relpath is None:
 
561
            raise AssertionError(
 
562
                "path for id {%s} is None!" % file_id)
 
563
        return self._kind(relpath)
 
564
 
 
565
    def _kind(self, relpath):
 
566
        abspath = self.abspath(relpath)
 
567
        kind = file_kind(abspath)
 
568
        if (self._repo_supports_tree_reference and
 
569
            kind == 'directory' and
 
570
            self._directory_is_tree_reference(relpath)):
 
571
            kind = 'tree-reference'
 
572
        return kind
 
573
 
 
574
    @needs_read_lock
 
575
    def _last_revision(self):
 
576
        """See Mutable.last_revision."""
 
577
        parent_ids = self.current_dirstate().get_parent_ids()
 
578
        if parent_ids:
 
579
            return parent_ids[0]
 
580
        else:
 
581
            return _mod_revision.NULL_REVISION
 
582
 
 
583
    def lock_read(self):
 
584
        """See Branch.lock_read, and WorkingTree.unlock."""
 
585
        self.branch.lock_read()
 
586
        try:
 
587
            self._control_files.lock_read()
 
588
            try:
 
589
                state = self.current_dirstate()
 
590
                if not state._lock_token:
 
591
                    state.lock_read()
 
592
                # set our support for tree references from the repository in
 
593
                # use.
 
594
                self._repo_supports_tree_reference = getattr(
 
595
                    self.branch.repository._format, "supports_tree_reference",
 
596
                    False)
 
597
            except:
 
598
                self._control_files.unlock()
 
599
                raise
 
600
        except:
 
601
            self.branch.unlock()
 
602
            raise
 
603
 
 
604
    def _lock_self_write(self):
 
605
        """This should be called after the branch is locked."""
 
606
        try:
 
607
            self._control_files.lock_write()
 
608
            try:
 
609
                state = self.current_dirstate()
 
610
                if not state._lock_token:
 
611
                    state.lock_write()
 
612
                # set our support for tree references from the repository in
 
613
                # use.
 
614
                self._repo_supports_tree_reference = getattr(
 
615
                    self.branch.repository._format, "supports_tree_reference",
 
616
                    False)
 
617
            except:
 
618
                self._control_files.unlock()
 
619
                raise
 
620
        except:
 
621
            self.branch.unlock()
 
622
            raise
 
623
 
 
624
    def lock_tree_write(self):
 
625
        """See MutableTree.lock_tree_write, and WorkingTree.unlock."""
 
626
        self.branch.lock_read()
 
627
        self._lock_self_write()
 
628
 
 
629
    def lock_write(self):
 
630
        """See MutableTree.lock_write, and WorkingTree.unlock."""
 
631
        self.branch.lock_write()
 
632
        self._lock_self_write()
 
633
 
 
634
    @needs_tree_write_lock
 
635
    def move(self, from_paths, to_dir, after=False):
 
636
        """See WorkingTree.move()."""
 
637
        result = []
 
638
        if not from_paths:
 
639
            return result
 
640
        state = self.current_dirstate()
 
641
        if isinstance(from_paths, basestring):
 
642
            raise ValueError()
 
643
        to_dir_utf8 = to_dir.encode('utf8')
 
644
        to_entry_dirname, to_basename = os.path.split(to_dir_utf8)
 
645
        id_index = state._get_id_index()
 
646
        # check destination directory
 
647
        # get the details for it
 
648
        to_entry_block_index, to_entry_entry_index, dir_present, entry_present = \
 
649
            state._get_block_entry_index(to_entry_dirname, to_basename, 0)
 
650
        if not entry_present:
 
651
            raise errors.BzrMoveFailedError('', to_dir,
 
652
                errors.NotVersionedError(to_dir))
 
653
        to_entry = state._dirblocks[to_entry_block_index][1][to_entry_entry_index]
 
654
        # get a handle on the block itself.
 
655
        to_block_index = state._ensure_block(
 
656
            to_entry_block_index, to_entry_entry_index, to_dir_utf8)
 
657
        to_block = state._dirblocks[to_block_index]
 
658
        to_abs = self.abspath(to_dir)
 
659
        if not isdir(to_abs):
 
660
            raise errors.BzrMoveFailedError('',to_dir,
 
661
                errors.NotADirectory(to_abs))
 
662
 
 
663
        if to_entry[1][0][0] != 'd':
 
664
            raise errors.BzrMoveFailedError('',to_dir,
 
665
                errors.NotADirectory(to_abs))
 
666
 
 
667
        if self._inventory is not None:
 
668
            update_inventory = True
 
669
            inv = self.inventory
 
670
            to_dir_id = to_entry[0][2]
 
671
            to_dir_ie = inv[to_dir_id]
 
672
        else:
 
673
            update_inventory = False
 
674
 
 
675
        rollbacks = []
 
676
        def move_one(old_entry, from_path_utf8, minikind, executable,
 
677
                     fingerprint, packed_stat, size,
 
678
                     to_block, to_key, to_path_utf8):
 
679
            state._make_absent(old_entry)
 
680
            from_key = old_entry[0]
 
681
            rollbacks.append(
 
682
                lambda:state.update_minimal(from_key,
 
683
                    minikind,
 
684
                    executable=executable,
 
685
                    fingerprint=fingerprint,
 
686
                    packed_stat=packed_stat,
 
687
                    size=size,
 
688
                    path_utf8=from_path_utf8))
 
689
            state.update_minimal(to_key,
 
690
                    minikind,
 
691
                    executable=executable,
 
692
                    fingerprint=fingerprint,
 
693
                    packed_stat=packed_stat,
 
694
                    size=size,
 
695
                    path_utf8=to_path_utf8)
 
696
            added_entry_index, _ = state._find_entry_index(to_key, to_block[1])
 
697
            new_entry = to_block[1][added_entry_index]
 
698
            rollbacks.append(lambda:state._make_absent(new_entry))
 
699
 
 
700
        for from_rel in from_paths:
 
701
            # from_rel is 'pathinroot/foo/bar'
 
702
            from_rel_utf8 = from_rel.encode('utf8')
 
703
            from_dirname, from_tail = osutils.split(from_rel)
 
704
            from_dirname, from_tail_utf8 = osutils.split(from_rel_utf8)
 
705
            from_entry = self._get_entry(path=from_rel)
 
706
            if from_entry == (None, None):
 
707
                raise errors.BzrMoveFailedError(from_rel,to_dir,
 
708
                    errors.NotVersionedError(path=str(from_rel)))
 
709
 
 
710
            from_id = from_entry[0][2]
 
711
            to_rel = pathjoin(to_dir, from_tail)
 
712
            to_rel_utf8 = pathjoin(to_dir_utf8, from_tail_utf8)
 
713
            item_to_entry = self._get_entry(path=to_rel)
 
714
            if item_to_entry != (None, None):
 
715
                raise errors.BzrMoveFailedError(from_rel, to_rel,
 
716
                    "Target is already versioned.")
 
717
 
 
718
            if from_rel == to_rel:
 
719
                raise errors.BzrMoveFailedError(from_rel, to_rel,
 
720
                    "Source and target are identical.")
 
721
 
 
722
            from_missing = not self.has_filename(from_rel)
 
723
            to_missing = not self.has_filename(to_rel)
 
724
            if after:
 
725
                move_file = False
 
726
            else:
 
727
                move_file = True
 
728
            if to_missing:
 
729
                if not move_file:
 
730
                    raise errors.BzrMoveFailedError(from_rel, to_rel,
 
731
                        errors.NoSuchFile(path=to_rel,
 
732
                        extra="New file has not been created yet"))
 
733
                elif from_missing:
 
734
                    # neither path exists
 
735
                    raise errors.BzrRenameFailedError(from_rel, to_rel,
 
736
                        errors.PathsDoNotExist(paths=(from_rel, to_rel)))
 
737
            else:
 
738
                if from_missing: # implicitly just update our path mapping
 
739
                    move_file = False
 
740
                elif not after:
 
741
                    raise errors.RenameFailedFilesExist(from_rel, to_rel)
 
742
 
 
743
            rollbacks = []
 
744
            def rollback_rename():
 
745
                """A single rename has failed, roll it back."""
 
746
                # roll back everything, even if we encounter trouble doing one
 
747
                # of them.
 
748
                #
 
749
                # TODO: at least log the other exceptions rather than just
 
750
                # losing them mbp 20070307
 
751
                exc_info = None
 
752
                for rollback in reversed(rollbacks):
 
753
                    try:
 
754
                        rollback()
 
755
                    except Exception, e:
 
756
                        exc_info = sys.exc_info()
 
757
                if exc_info:
 
758
                    raise exc_info[0], exc_info[1], exc_info[2]
 
759
 
 
760
            # perform the disk move first - its the most likely failure point.
 
761
            if move_file:
 
762
                from_rel_abs = self.abspath(from_rel)
 
763
                to_rel_abs = self.abspath(to_rel)
 
764
                try:
 
765
                    osutils.rename(from_rel_abs, to_rel_abs)
 
766
                except OSError, e:
 
767
                    raise errors.BzrMoveFailedError(from_rel, to_rel, e[1])
 
768
                rollbacks.append(lambda: osutils.rename(to_rel_abs, from_rel_abs))
 
769
            try:
 
770
                # perform the rename in the inventory next if needed: its easy
 
771
                # to rollback
 
772
                if update_inventory:
 
773
                    # rename the entry
 
774
                    from_entry = inv[from_id]
 
775
                    current_parent = from_entry.parent_id
 
776
                    inv.rename(from_id, to_dir_id, from_tail)
 
777
                    rollbacks.append(
 
778
                        lambda: inv.rename(from_id, current_parent, from_tail))
 
779
                # finally do the rename in the dirstate, which is a little
 
780
                # tricky to rollback, but least likely to need it.
 
781
                old_block_index, old_entry_index, dir_present, file_present = \
 
782
                    state._get_block_entry_index(from_dirname, from_tail_utf8, 0)
 
783
                old_block = state._dirblocks[old_block_index][1]
 
784
                old_entry = old_block[old_entry_index]
 
785
                from_key, old_entry_details = old_entry
 
786
                cur_details = old_entry_details[0]
 
787
                # remove the old row
 
788
                to_key = ((to_block[0],) + from_key[1:3])
 
789
                minikind = cur_details[0]
 
790
                move_one(old_entry, from_path_utf8=from_rel_utf8,
 
791
                         minikind=minikind,
 
792
                         executable=cur_details[3],
 
793
                         fingerprint=cur_details[1],
 
794
                         packed_stat=cur_details[4],
 
795
                         size=cur_details[2],
 
796
                         to_block=to_block,
 
797
                         to_key=to_key,
 
798
                         to_path_utf8=to_rel_utf8)
 
799
 
 
800
                if minikind == 'd':
 
801
                    def update_dirblock(from_dir, to_key, to_dir_utf8):
 
802
                        """Recursively update all entries in this dirblock."""
 
803
                        if from_dir == '':
 
804
                            raise AssertionError("renaming root not supported")
 
805
                        from_key = (from_dir, '')
 
806
                        from_block_idx, present = \
 
807
                            state._find_block_index_from_key(from_key)
 
808
                        if not present:
 
809
                            # This is the old record, if it isn't present, then
 
810
                            # there is theoretically nothing to update.
 
811
                            # (Unless it isn't present because of lazy loading,
 
812
                            # but we don't do that yet)
 
813
                            return
 
814
                        from_block = state._dirblocks[from_block_idx]
 
815
                        to_block_index, to_entry_index, _, _ = \
 
816
                            state._get_block_entry_index(to_key[0], to_key[1], 0)
 
817
                        to_block_index = state._ensure_block(
 
818
                            to_block_index, to_entry_index, to_dir_utf8)
 
819
                        to_block = state._dirblocks[to_block_index]
 
820
 
 
821
                        # Grab a copy since move_one may update the list.
 
822
                        for entry in from_block[1][:]:
 
823
                            if not (entry[0][0] == from_dir):
 
824
                                raise AssertionError()
 
825
                            cur_details = entry[1][0]
 
826
                            to_key = (to_dir_utf8, entry[0][1], entry[0][2])
 
827
                            from_path_utf8 = osutils.pathjoin(entry[0][0], entry[0][1])
 
828
                            to_path_utf8 = osutils.pathjoin(to_dir_utf8, entry[0][1])
 
829
                            minikind = cur_details[0]
 
830
                            if minikind in 'ar':
 
831
                                # Deleted children of a renamed directory
 
832
                                # Do not need to be updated.
 
833
                                # Children that have been renamed out of this
 
834
                                # directory should also not be updated
 
835
                                continue
 
836
                            move_one(entry, from_path_utf8=from_path_utf8,
 
837
                                     minikind=minikind,
 
838
                                     executable=cur_details[3],
 
839
                                     fingerprint=cur_details[1],
 
840
                                     packed_stat=cur_details[4],
 
841
                                     size=cur_details[2],
 
842
                                     to_block=to_block,
 
843
                                     to_key=to_key,
 
844
                                     to_path_utf8=to_path_utf8)
 
845
                            if minikind == 'd':
 
846
                                # We need to move all the children of this
 
847
                                # entry
 
848
                                update_dirblock(from_path_utf8, to_key,
 
849
                                                to_path_utf8)
 
850
                    update_dirblock(from_rel_utf8, to_key, to_rel_utf8)
 
851
            except:
 
852
                rollback_rename()
 
853
                raise
 
854
            result.append((from_rel, to_rel))
 
855
            state._dirblock_state = dirstate.DirState.IN_MEMORY_MODIFIED
 
856
            self._make_dirty(reset_inventory=False)
 
857
 
 
858
        return result
 
859
 
 
860
    def _must_be_locked(self):
 
861
        if not self._control_files._lock_count:
 
862
            raise errors.ObjectNotLocked(self)
 
863
 
 
864
    def _new_tree(self):
 
865
        """Initialize the state in this tree to be a new tree."""
 
866
        self._dirty = True
 
867
 
 
868
    @needs_read_lock
 
869
    def path2id(self, path):
 
870
        """Return the id for path in this tree."""
 
871
        path = path.strip('/')
 
872
        entry = self._get_entry(path=path)
 
873
        if entry == (None, None):
 
874
            return None
 
875
        return entry[0][2]
 
876
 
 
877
    def paths2ids(self, paths, trees=[], require_versioned=True):
 
878
        """See Tree.paths2ids().
 
879
 
 
880
        This specialisation fast-paths the case where all the trees are in the
 
881
        dirstate.
 
882
        """
 
883
        if paths is None:
 
884
            return None
 
885
        parents = self.get_parent_ids()
 
886
        for tree in trees:
 
887
            if not (isinstance(tree, DirStateRevisionTree) and tree._revision_id in
 
888
                parents):
 
889
                return super(WorkingTree4, self).paths2ids(paths, trees, require_versioned)
 
890
        search_indexes = [0] + [1 + parents.index(tree._revision_id) for tree in trees]
 
891
        # -- make all paths utf8 --
 
892
        paths_utf8 = set()
 
893
        for path in paths:
 
894
            paths_utf8.add(path.encode('utf8'))
 
895
        paths = paths_utf8
 
896
        # -- paths is now a utf8 path set --
 
897
        # -- get the state object and prepare it.
 
898
        state = self.current_dirstate()
 
899
        if False and (state._dirblock_state == dirstate.DirState.NOT_IN_MEMORY
 
900
            and '' not in paths):
 
901
            paths2ids = self._paths2ids_using_bisect
 
902
        else:
 
903
            paths2ids = self._paths2ids_in_memory
 
904
        return paths2ids(paths, search_indexes,
 
905
                         require_versioned=require_versioned)
 
906
 
 
907
    def _paths2ids_in_memory(self, paths, search_indexes,
 
908
                             require_versioned=True):
 
909
        state = self.current_dirstate()
 
910
        state._read_dirblocks_if_needed()
 
911
        def _entries_for_path(path):
 
912
            """Return a list with all the entries that match path for all ids.
 
913
            """
 
914
            dirname, basename = os.path.split(path)
 
915
            key = (dirname, basename, '')
 
916
            block_index, present = state._find_block_index_from_key(key)
 
917
            if not present:
 
918
                # the block which should contain path is absent.
 
919
                return []
 
920
            result = []
 
921
            block = state._dirblocks[block_index][1]
 
922
            entry_index, _ = state._find_entry_index(key, block)
 
923
            # we may need to look at multiple entries at this path: walk while the paths match.
 
924
            while (entry_index < len(block) and
 
925
                block[entry_index][0][0:2] == key[0:2]):
 
926
                result.append(block[entry_index])
 
927
                entry_index += 1
 
928
            return result
 
929
        if require_versioned:
 
930
            # -- check all supplied paths are versioned in a search tree. --
 
931
            all_versioned = True
 
932
            for path in paths:
 
933
                path_entries = _entries_for_path(path)
 
934
                if not path_entries:
 
935
                    # this specified path is not present at all: error
 
936
                    all_versioned = False
 
937
                    break
 
938
                found_versioned = False
 
939
                # for each id at this path
 
940
                for entry in path_entries:
 
941
                    # for each tree.
 
942
                    for index in search_indexes:
 
943
                        if entry[1][index][0] != 'a': # absent
 
944
                            found_versioned = True
 
945
                            # all good: found a versioned cell
 
946
                            break
 
947
                if not found_versioned:
 
948
                    # none of the indexes was not 'absent' at all ids for this
 
949
                    # path.
 
950
                    all_versioned = False
 
951
                    break
 
952
            if not all_versioned:
 
953
                raise errors.PathsNotVersionedError(paths)
 
954
        # -- remove redundancy in supplied paths to prevent over-scanning --
 
955
        search_paths = osutils.minimum_path_selection(paths)
 
956
        # sketch: 
 
957
        # for all search_indexs in each path at or under each element of
 
958
        # search_paths, if the detail is relocated: add the id, and add the
 
959
        # relocated path as one to search if its not searched already. If the
 
960
        # detail is not relocated, add the id.
 
961
        searched_paths = set()
 
962
        found_ids = set()
 
963
        def _process_entry(entry):
 
964
            """Look at search_indexes within entry.
 
965
 
 
966
            If a specific tree's details are relocated, add the relocation
 
967
            target to search_paths if not searched already. If it is absent, do
 
968
            nothing. Otherwise add the id to found_ids.
 
969
            """
 
970
            for index in search_indexes:
 
971
                if entry[1][index][0] == 'r': # relocated
 
972
                    if not osutils.is_inside_any(searched_paths, entry[1][index][1]):
 
973
                        search_paths.add(entry[1][index][1])
 
974
                elif entry[1][index][0] != 'a': # absent
 
975
                    found_ids.add(entry[0][2])
 
976
        while search_paths:
 
977
            current_root = search_paths.pop()
 
978
            searched_paths.add(current_root)
 
979
            # process the entries for this containing directory: the rest will be
 
980
            # found by their parents recursively.
 
981
            root_entries = _entries_for_path(current_root)
 
982
            if not root_entries:
 
983
                # this specified path is not present at all, skip it.
 
984
                continue
 
985
            for entry in root_entries:
 
986
                _process_entry(entry)
 
987
            initial_key = (current_root, '', '')
 
988
            block_index, _ = state._find_block_index_from_key(initial_key)
 
989
            while (block_index < len(state._dirblocks) and
 
990
                osutils.is_inside(current_root, state._dirblocks[block_index][0])):
 
991
                for entry in state._dirblocks[block_index][1]:
 
992
                    _process_entry(entry)
 
993
                block_index += 1
 
994
        return found_ids
 
995
 
 
996
    def _paths2ids_using_bisect(self, paths, search_indexes,
 
997
                                require_versioned=True):
 
998
        state = self.current_dirstate()
 
999
        found_ids = set()
 
1000
 
 
1001
        split_paths = sorted(osutils.split(p) for p in paths)
 
1002
        found = state._bisect_recursive(split_paths)
 
1003
 
 
1004
        if require_versioned:
 
1005
            found_dir_names = set(dir_name_id[:2] for dir_name_id in found)
 
1006
            for dir_name in split_paths:
 
1007
                if dir_name not in found_dir_names:
 
1008
                    raise errors.PathsNotVersionedError(paths)
 
1009
 
 
1010
        for dir_name_id, trees_info in found.iteritems():
 
1011
            for index in search_indexes:
 
1012
                if trees_info[index][0] not in ('r', 'a'):
 
1013
                    found_ids.add(dir_name_id[2])
 
1014
        return found_ids
 
1015
 
 
1016
    def read_working_inventory(self):
 
1017
        """Read the working inventory.
 
1018
        
 
1019
        This is a meaningless operation for dirstate, but we obey it anyhow.
 
1020
        """
 
1021
        return self.inventory
 
1022
 
 
1023
    @needs_read_lock
 
1024
    def revision_tree(self, revision_id):
 
1025
        """See Tree.revision_tree.
 
1026
 
 
1027
        WorkingTree4 supplies revision_trees for any basis tree.
 
1028
        """
 
1029
        dirstate = self.current_dirstate()
 
1030
        parent_ids = dirstate.get_parent_ids()
 
1031
        if revision_id not in parent_ids:
 
1032
            raise errors.NoSuchRevisionInTree(self, revision_id)
 
1033
        if revision_id in dirstate.get_ghosts():
 
1034
            raise errors.NoSuchRevisionInTree(self, revision_id)
 
1035
        return DirStateRevisionTree(dirstate, revision_id,
 
1036
            self.branch.repository)
 
1037
 
 
1038
    @needs_tree_write_lock
 
1039
    def set_last_revision(self, new_revision):
 
1040
        """Change the last revision in the working tree."""
 
1041
        parents = self.get_parent_ids()
 
1042
        if new_revision in (NULL_REVISION, None):
 
1043
            if len(parents) >= 2:
 
1044
                raise AssertionError(
 
1045
                    "setting the last parent to none with a pending merge is "
 
1046
                    "unsupported.")
 
1047
            self.set_parent_ids([])
 
1048
        else:
 
1049
            self.set_parent_ids([new_revision] + parents[1:],
 
1050
                allow_leftmost_as_ghost=True)
 
1051
 
 
1052
    @needs_tree_write_lock
 
1053
    def set_parent_ids(self, revision_ids, allow_leftmost_as_ghost=False):
 
1054
        """Set the parent ids to revision_ids.
 
1055
        
 
1056
        See also set_parent_trees. This api will try to retrieve the tree data
 
1057
        for each element of revision_ids from the trees repository. If you have
 
1058
        tree data already available, it is more efficient to use
 
1059
        set_parent_trees rather than set_parent_ids. set_parent_ids is however
 
1060
        an easier API to use.
 
1061
 
 
1062
        :param revision_ids: The revision_ids to set as the parent ids of this
 
1063
            working tree. Any of these may be ghosts.
 
1064
        """
 
1065
        trees = []
 
1066
        for revision_id in revision_ids:
 
1067
            try:
 
1068
                revtree = self.branch.repository.revision_tree(revision_id)
 
1069
                # TODO: jam 20070213 KnitVersionedFile raises
 
1070
                #       RevisionNotPresent rather than NoSuchRevision if a
 
1071
                #       given revision_id is not present. Should Repository be
 
1072
                #       catching it and re-raising NoSuchRevision?
 
1073
            except (errors.NoSuchRevision, errors.RevisionNotPresent):
 
1074
                revtree = None
 
1075
            trees.append((revision_id, revtree))
 
1076
        self.set_parent_trees(trees,
 
1077
            allow_leftmost_as_ghost=allow_leftmost_as_ghost)
 
1078
 
 
1079
    @needs_tree_write_lock
 
1080
    def set_parent_trees(self, parents_list, allow_leftmost_as_ghost=False):
 
1081
        """Set the parents of the working tree.
 
1082
 
 
1083
        :param parents_list: A list of (revision_id, tree) tuples.
 
1084
            If tree is None, then that element is treated as an unreachable
 
1085
            parent tree - i.e. a ghost.
 
1086
        """
 
1087
        dirstate = self.current_dirstate()
 
1088
        if len(parents_list) > 0:
 
1089
            if not allow_leftmost_as_ghost and parents_list[0][1] is None:
 
1090
                raise errors.GhostRevisionUnusableHere(parents_list[0][0])
 
1091
        real_trees = []
 
1092
        ghosts = []
 
1093
 
 
1094
        parent_ids = [rev_id for rev_id, tree in parents_list]
 
1095
        graph = self.branch.repository.get_graph()
 
1096
        heads = graph.heads(parent_ids)
 
1097
        accepted_revisions = set()
 
1098
 
 
1099
        # convert absent trees to the null tree, which we convert back to
 
1100
        # missing on access.
 
1101
        for rev_id, tree in parents_list:
 
1102
            if len(accepted_revisions) > 0:
 
1103
                # we always accept the first tree
 
1104
                if rev_id in accepted_revisions or rev_id not in heads:
 
1105
                    # We have already included either this tree, or its
 
1106
                    # descendent, so we skip it.
 
1107
                    continue
 
1108
            _mod_revision.check_not_reserved_id(rev_id)
 
1109
            if tree is not None:
 
1110
                real_trees.append((rev_id, tree))
 
1111
            else:
 
1112
                real_trees.append((rev_id,
 
1113
                    self.branch.repository.revision_tree(None)))
 
1114
                ghosts.append(rev_id)
 
1115
            accepted_revisions.add(rev_id)
 
1116
        dirstate.set_parent_trees(real_trees, ghosts=ghosts)
 
1117
        self._make_dirty(reset_inventory=False)
 
1118
 
 
1119
    def _set_root_id(self, file_id):
 
1120
        """See WorkingTree.set_root_id."""
 
1121
        state = self.current_dirstate()
 
1122
        state.set_path_id('', file_id)
 
1123
        if state._dirblock_state == dirstate.DirState.IN_MEMORY_MODIFIED:
 
1124
            self._make_dirty(reset_inventory=True)
 
1125
 
 
1126
    def _sha_from_stat(self, path, stat_result):
 
1127
        """Get a sha digest from the tree's stat cache.
 
1128
 
 
1129
        The default implementation assumes no stat cache is present.
 
1130
 
 
1131
        :param path: The path.
 
1132
        :param stat_result: The stat result being looked up.
 
1133
        """
 
1134
        return self.current_dirstate().sha1_from_stat(path, stat_result)
 
1135
 
 
1136
    @needs_read_lock
 
1137
    def supports_tree_reference(self):
 
1138
        return self._repo_supports_tree_reference
 
1139
 
 
1140
    def unlock(self):
 
1141
        """Unlock in format 4 trees needs to write the entire dirstate."""
 
1142
        # do non-implementation specific cleanup
 
1143
        self._cleanup()
 
1144
 
 
1145
        if self._control_files._lock_count == 1:
 
1146
            # eventually we should do signature checking during read locks for
 
1147
            # dirstate updates.
 
1148
            if self._control_files._lock_mode == 'w':
 
1149
                if self._dirty:
 
1150
                    self.flush()
 
1151
            if self._dirstate is not None:
 
1152
                # This is a no-op if there are no modifications.
 
1153
                self._dirstate.save()
 
1154
                self._dirstate.unlock()
 
1155
            # TODO: jam 20070301 We shouldn't have to wipe the dirstate at this
 
1156
            #       point. Instead, it could check if the header has been
 
1157
            #       modified when it is locked, and if not, it can hang on to
 
1158
            #       the data it has in memory.
 
1159
            self._dirstate = None
 
1160
            self._inventory = None
 
1161
        # reverse order of locking.
 
1162
        try:
 
1163
            return self._control_files.unlock()
 
1164
        finally:
 
1165
            self.branch.unlock()
 
1166
 
 
1167
    @needs_tree_write_lock
 
1168
    def unversion(self, file_ids):
 
1169
        """Remove the file ids in file_ids from the current versioned set.
 
1170
 
 
1171
        When a file_id is unversioned, all of its children are automatically
 
1172
        unversioned.
 
1173
 
 
1174
        :param file_ids: The file ids to stop versioning.
 
1175
        :raises: NoSuchId if any fileid is not currently versioned.
 
1176
        """
 
1177
        if not file_ids:
 
1178
            return
 
1179
        state = self.current_dirstate()
 
1180
        state._read_dirblocks_if_needed()
 
1181
        ids_to_unversion = set(file_ids)
 
1182
        paths_to_unversion = set()
 
1183
        # sketch:
 
1184
        # check if the root is to be unversioned, if so, assert for now.
 
1185
        # walk the state marking unversioned things as absent.
 
1186
        # if there are any un-unversioned ids at the end, raise
 
1187
        for key, details in state._dirblocks[0][1]:
 
1188
            if (details[0][0] not in ('a', 'r') and # absent or relocated
 
1189
                key[2] in ids_to_unversion):
 
1190
                # I haven't written the code to unversion / yet - it should be
 
1191
                # supported.
 
1192
                raise errors.BzrError('Unversioning the / is not currently supported')
 
1193
        block_index = 0
 
1194
        while block_index < len(state._dirblocks):
 
1195
            # process one directory at a time.
 
1196
            block = state._dirblocks[block_index]
 
1197
            # first check: is the path one to remove - it or its children
 
1198
            delete_block = False
 
1199
            for path in paths_to_unversion:
 
1200
                if (block[0].startswith(path) and
 
1201
                    (len(block[0]) == len(path) or
 
1202
                     block[0][len(path)] == '/')):
 
1203
                    # this entire block should be deleted - its the block for a
 
1204
                    # path to unversion; or the child of one
 
1205
                    delete_block = True
 
1206
                    break
 
1207
            # TODO: trim paths_to_unversion as we pass by paths
 
1208
            if delete_block:
 
1209
                # this block is to be deleted: process it.
 
1210
                # TODO: we can special case the no-parents case and
 
1211
                # just forget the whole block.
 
1212
                entry_index = 0
 
1213
                while entry_index < len(block[1]):
 
1214
                    # Mark this file id as having been removed
 
1215
                    entry = block[1][entry_index]
 
1216
                    ids_to_unversion.discard(entry[0][2])
 
1217
                    if (entry[1][0][0] in 'ar' # don't remove absent or renamed
 
1218
                                               # entries
 
1219
                        or not state._make_absent(entry)):
 
1220
                        entry_index += 1
 
1221
                # go to the next block. (At the moment we dont delete empty
 
1222
                # dirblocks)
 
1223
                block_index += 1
 
1224
                continue
 
1225
            entry_index = 0
 
1226
            while entry_index < len(block[1]):
 
1227
                entry = block[1][entry_index]
 
1228
                if (entry[1][0][0] in ('a', 'r') or # absent, relocated
 
1229
                    # ^ some parent row.
 
1230
                    entry[0][2] not in ids_to_unversion):
 
1231
                    # ^ not an id to unversion
 
1232
                    entry_index += 1
 
1233
                    continue
 
1234
                if entry[1][0][0] == 'd':
 
1235
                    paths_to_unversion.add(pathjoin(entry[0][0], entry[0][1]))
 
1236
                if not state._make_absent(entry):
 
1237
                    entry_index += 1
 
1238
                # we have unversioned this id
 
1239
                ids_to_unversion.remove(entry[0][2])
 
1240
            block_index += 1
 
1241
        if ids_to_unversion:
 
1242
            raise errors.NoSuchId(self, iter(ids_to_unversion).next())
 
1243
        self._make_dirty(reset_inventory=False)
 
1244
        # have to change the legacy inventory too.
 
1245
        if self._inventory is not None:
 
1246
            for file_id in file_ids:
 
1247
                self._inventory.remove_recursive_id(file_id)
 
1248
 
 
1249
    @needs_tree_write_lock
 
1250
    def rename_one(self, from_rel, to_rel, after=False):
 
1251
        """See WorkingTree.rename_one"""
 
1252
        self.flush()
 
1253
        WorkingTree.rename_one(self, from_rel, to_rel, after)
 
1254
 
 
1255
    @needs_tree_write_lock
 
1256
    def apply_inventory_delta(self, changes):
 
1257
        """See MutableTree.apply_inventory_delta"""
 
1258
        state = self.current_dirstate()
 
1259
        state.update_by_delta(changes)
 
1260
        self._make_dirty(reset_inventory=True)
 
1261
 
 
1262
    def update_basis_by_delta(self, new_revid, delta):
 
1263
        """See MutableTree.update_basis_by_delta."""
 
1264
        if self.last_revision() == new_revid:
 
1265
            raise AssertionError()
 
1266
        self.current_dirstate().update_basis_by_delta(delta, new_revid)
 
1267
 
 
1268
    @needs_read_lock
 
1269
    def _validate(self):
 
1270
        self._dirstate._validate()
 
1271
 
 
1272
    @needs_tree_write_lock
 
1273
    def _write_inventory(self, inv):
 
1274
        """Write inventory as the current inventory."""
 
1275
        if self._dirty:
 
1276
            raise AssertionError("attempting to write an inventory when the "
 
1277
                "dirstate is dirty will lose pending changes")
 
1278
        self.current_dirstate().set_state_from_inventory(inv)
 
1279
        self._make_dirty(reset_inventory=False)
 
1280
        if self._inventory is not None:
 
1281
            self._inventory = inv
 
1282
        self.flush()
 
1283
 
 
1284
 
 
1285
class WorkingTreeFormat4(WorkingTreeFormat3):
 
1286
    """The first consolidated dirstate working tree format.
 
1287
 
 
1288
    This format:
 
1289
        - exists within a metadir controlling .bzr
 
1290
        - includes an explicit version marker for the workingtree control
 
1291
          files, separate from the BzrDir format
 
1292
        - modifies the hash cache format
 
1293
        - is new in bzr 0.15
 
1294
        - uses a LockDir to guard access to it.
 
1295
    """
 
1296
 
 
1297
    upgrade_recommended = False
 
1298
 
 
1299
    def get_format_string(self):
 
1300
        """See WorkingTreeFormat.get_format_string()."""
 
1301
        return "Bazaar Working Tree Format 4 (bzr 0.15)\n"
 
1302
 
 
1303
    def get_format_description(self):
 
1304
        """See WorkingTreeFormat.get_format_description()."""
 
1305
        return "Working tree format 4"
 
1306
 
 
1307
    def initialize(self, a_bzrdir, revision_id=None, from_branch=None,
 
1308
                   accelerator_tree=None, hardlink=False):
 
1309
        """See WorkingTreeFormat.initialize().
 
1310
 
 
1311
        :param revision_id: allows creating a working tree at a different
 
1312
        revision than the branch is at.
 
1313
        :param accelerator_tree: A tree which can be used for retrieving file
 
1314
            contents more quickly than the revision tree, i.e. a workingtree.
 
1315
            The revision tree will be used for cases where accelerator_tree's
 
1316
            content is different.
 
1317
        :param hardlink: If true, hard-link files from accelerator_tree,
 
1318
            where possible.
 
1319
 
 
1320
        These trees get an initial random root id, if their repository supports
 
1321
        rich root data, TREE_ROOT otherwise.
 
1322
        """
 
1323
        if not isinstance(a_bzrdir.transport, LocalTransport):
 
1324
            raise errors.NotLocalUrl(a_bzrdir.transport.base)
 
1325
        transport = a_bzrdir.get_workingtree_transport(self)
 
1326
        control_files = self._open_control_files(a_bzrdir)
 
1327
        control_files.create_lock()
 
1328
        control_files.lock_write()
 
1329
        transport.put_bytes('format', self.get_format_string(),
 
1330
            mode=a_bzrdir._get_file_mode())
 
1331
        if from_branch is not None:
 
1332
            branch = from_branch
 
1333
        else:
 
1334
            branch = a_bzrdir.open_branch()
 
1335
        if revision_id is None:
 
1336
            revision_id = branch.last_revision()
 
1337
        local_path = transport.local_abspath('dirstate')
 
1338
        # write out new dirstate (must exist when we create the tree)
 
1339
        state = dirstate.DirState.initialize(local_path)
 
1340
        state.unlock()
 
1341
        del state
 
1342
        wt = WorkingTree4(a_bzrdir.root_transport.local_abspath('.'),
 
1343
                         branch,
 
1344
                         _format=self,
 
1345
                         _bzrdir=a_bzrdir,
 
1346
                         _control_files=control_files)
 
1347
        wt._new_tree()
 
1348
        wt.lock_tree_write()
 
1349
        try:
 
1350
            if revision_id in (None, NULL_REVISION):
 
1351
                if branch.repository.supports_rich_root():
 
1352
                    wt._set_root_id(generate_ids.gen_root_id())
 
1353
                else:
 
1354
                    wt._set_root_id(ROOT_ID)
 
1355
                wt.flush()
 
1356
            basis = None
 
1357
            # frequently, we will get here due to branching.  The accelerator
 
1358
            # tree will be the tree from the branch, so the desired basis
 
1359
            # tree will often be a parent of the accelerator tree.
 
1360
            if accelerator_tree is not None:
 
1361
                try:
 
1362
                    basis = accelerator_tree.revision_tree(revision_id)
 
1363
                except errors.NoSuchRevision:
 
1364
                    pass
 
1365
            if basis is None:
 
1366
                basis = branch.repository.revision_tree(revision_id)
 
1367
            if revision_id == NULL_REVISION:
 
1368
                parents_list = []
 
1369
            else:
 
1370
                parents_list = [(revision_id, basis)]
 
1371
            basis.lock_read()
 
1372
            try:
 
1373
                wt.set_parent_trees(parents_list, allow_leftmost_as_ghost=True)
 
1374
                wt.flush()
 
1375
                # if the basis has a root id we have to use that; otherwise we
 
1376
                # use a new random one
 
1377
                basis_root_id = basis.get_root_id()
 
1378
                if basis_root_id is not None:
 
1379
                    wt._set_root_id(basis_root_id)
 
1380
                    wt.flush()
 
1381
                # delta_from_tree is safe even for DirStateRevisionTrees,
 
1382
                # because wt4.apply_inventory_delta does not mutate the input
 
1383
                # inventory entries.
 
1384
                transform.build_tree(basis, wt, accelerator_tree,
 
1385
                                     hardlink=hardlink, delta_from_tree=True)
 
1386
            finally:
 
1387
                basis.unlock()
 
1388
        finally:
 
1389
            control_files.unlock()
 
1390
            wt.unlock()
 
1391
        return wt
 
1392
 
 
1393
    def _open(self, a_bzrdir, control_files):
 
1394
        """Open the tree itself.
 
1395
 
 
1396
        :param a_bzrdir: the dir for the tree.
 
1397
        :param control_files: the control files for the tree.
 
1398
        """
 
1399
        return WorkingTree4(a_bzrdir.root_transport.local_abspath('.'),
 
1400
                           branch=a_bzrdir.open_branch(),
 
1401
                           _format=self,
 
1402
                           _bzrdir=a_bzrdir,
 
1403
                           _control_files=control_files)
 
1404
 
 
1405
    def __get_matchingbzrdir(self):
 
1406
        # please test against something that will let us do tree references
 
1407
        return bzrdir.format_registry.make_bzrdir(
 
1408
            'dirstate-with-subtree')
 
1409
 
 
1410
    _matchingbzrdir = property(__get_matchingbzrdir)
 
1411
 
 
1412
 
 
1413
class DirStateRevisionTree(Tree):
 
1414
    """A revision tree pulling the inventory from a dirstate."""
 
1415
 
 
1416
    def __init__(self, dirstate, revision_id, repository):
 
1417
        self._dirstate = dirstate
 
1418
        self._revision_id = revision_id
 
1419
        self._repository = repository
 
1420
        self._inventory = None
 
1421
        self._locked = 0
 
1422
        self._dirstate_locked = False
 
1423
        self._repo_supports_tree_reference = getattr(
 
1424
            repository._format, "supports_tree_reference",
 
1425
            False)
 
1426
 
 
1427
    def __repr__(self):
 
1428
        return "<%s of %s in %s>" % \
 
1429
            (self.__class__.__name__, self._revision_id, self._dirstate)
 
1430
 
 
1431
    def annotate_iter(self, file_id,
 
1432
                      default_revision=_mod_revision.CURRENT_REVISION):
 
1433
        """See Tree.annotate_iter"""
 
1434
        text_key = (file_id, self.inventory[file_id].revision)
 
1435
        annotations = self._repository.texts.annotate(text_key)
 
1436
        return [(key[-1], line) for (key, line) in annotations]
 
1437
 
 
1438
    def _get_ancestors(self, default_revision):
 
1439
        return set(self._repository.get_ancestry(self._revision_id,
 
1440
                                                 topo_sorted=False))
 
1441
    def _comparison_data(self, entry, path):
 
1442
        """See Tree._comparison_data."""
 
1443
        if entry is None:
 
1444
            return None, False, None
 
1445
        # trust the entry as RevisionTree does, but this may not be
 
1446
        # sensible: the entry might not have come from us?
 
1447
        return entry.kind, entry.executable, None
 
1448
 
 
1449
    def _file_size(self, entry, stat_value):
 
1450
        return entry.text_size
 
1451
 
 
1452
    def filter_unversioned_files(self, paths):
 
1453
        """Filter out paths that are not versioned.
 
1454
 
 
1455
        :return: set of paths.
 
1456
        """
 
1457
        pred = self.has_filename
 
1458
        return set((p for p in paths if not pred(p)))
 
1459
 
 
1460
    def get_root_id(self):
 
1461
        return self.path2id('')
 
1462
 
 
1463
    def id2path(self, file_id):
 
1464
        "Convert a file-id to a path."
 
1465
        entry = self._get_entry(file_id=file_id)
 
1466
        if entry == (None, None):
 
1467
            raise errors.NoSuchId(tree=self, file_id=file_id)
 
1468
        path_utf8 = osutils.pathjoin(entry[0][0], entry[0][1])
 
1469
        return path_utf8.decode('utf8')
 
1470
 
 
1471
    def iter_references(self):
 
1472
        if not self._repo_supports_tree_reference:
 
1473
            # When the repo doesn't support references, we will have nothing to
 
1474
            # return
 
1475
            return iter([])
 
1476
        # Otherwise, fall back to the default implementation
 
1477
        return super(DirStateRevisionTree, self).iter_references()
 
1478
 
 
1479
    def _get_parent_index(self):
 
1480
        """Return the index in the dirstate referenced by this tree."""
 
1481
        return self._dirstate.get_parent_ids().index(self._revision_id) + 1
 
1482
 
 
1483
    def _get_entry(self, file_id=None, path=None):
 
1484
        """Get the dirstate row for file_id or path.
 
1485
 
 
1486
        If either file_id or path is supplied, it is used as the key to lookup.
 
1487
        If both are supplied, the fastest lookup is used, and an error is
 
1488
        raised if they do not both point at the same row.
 
1489
        
 
1490
        :param file_id: An optional unicode file_id to be looked up.
 
1491
        :param path: An optional unicode path to be looked up.
 
1492
        :return: The dirstate row tuple for path/file_id, or (None, None)
 
1493
        """
 
1494
        if file_id is None and path is None:
 
1495
            raise errors.BzrError('must supply file_id or path')
 
1496
        if path is not None:
 
1497
            path = path.encode('utf8')
 
1498
        parent_index = self._get_parent_index()
 
1499
        return self._dirstate._get_entry(parent_index, fileid_utf8=file_id, path_utf8=path)
 
1500
 
 
1501
    def _generate_inventory(self):
 
1502
        """Create and set self.inventory from the dirstate object.
 
1503
 
 
1504
        (So this is only called the first time the inventory is requested for
 
1505
        this tree; it then remains in memory until it's out of date.)
 
1506
 
 
1507
        This is relatively expensive: we have to walk the entire dirstate.
 
1508
        """
 
1509
        if not self._locked:
 
1510
            raise AssertionError(
 
1511
                'cannot generate inventory of an unlocked '
 
1512
                'dirstate revision tree')
 
1513
        # separate call for profiling - makes it clear where the costs are.
 
1514
        self._dirstate._read_dirblocks_if_needed()
 
1515
        if self._revision_id not in self._dirstate.get_parent_ids():
 
1516
            raise AssertionError(
 
1517
                'parent %s has disappeared from %s' % (
 
1518
                self._revision_id, self._dirstate.get_parent_ids()))
 
1519
        parent_index = self._dirstate.get_parent_ids().index(self._revision_id) + 1
 
1520
        # This is identical now to the WorkingTree _generate_inventory except
 
1521
        # for the tree index use.
 
1522
        root_key, current_entry = self._dirstate._get_entry(parent_index, path_utf8='')
 
1523
        current_id = root_key[2]
 
1524
        if current_entry[parent_index][0] != 'd':
 
1525
            raise AssertionError()
 
1526
        inv = Inventory(root_id=current_id, revision_id=self._revision_id)
 
1527
        inv.root.revision = current_entry[parent_index][4]
 
1528
        # Turn some things into local variables
 
1529
        minikind_to_kind = dirstate.DirState._minikind_to_kind
 
1530
        factory = entry_factory
 
1531
        utf8_decode = cache_utf8._utf8_decode
 
1532
        inv_byid = inv._byid
 
1533
        # we could do this straight out of the dirstate; it might be fast
 
1534
        # and should be profiled - RBC 20070216
 
1535
        parent_ies = {'' : inv.root}
 
1536
        for block in self._dirstate._dirblocks[1:]: #skip root
 
1537
            dirname = block[0]
 
1538
            try:
 
1539
                parent_ie = parent_ies[dirname]
 
1540
            except KeyError:
 
1541
                # all the paths in this block are not versioned in this tree
 
1542
                continue
 
1543
            for key, entry in block[1]:
 
1544
                minikind, fingerprint, size, executable, revid = entry[parent_index]
 
1545
                if minikind in ('a', 'r'): # absent, relocated
 
1546
                    # not this tree
 
1547
                    continue
 
1548
                name = key[1]
 
1549
                name_unicode = utf8_decode(name)[0]
 
1550
                file_id = key[2]
 
1551
                kind = minikind_to_kind[minikind]
 
1552
                inv_entry = factory[kind](file_id, name_unicode,
 
1553
                                          parent_ie.file_id)
 
1554
                inv_entry.revision = revid
 
1555
                if kind == 'file':
 
1556
                    inv_entry.executable = executable
 
1557
                    inv_entry.text_size = size
 
1558
                    inv_entry.text_sha1 = fingerprint
 
1559
                elif kind == 'directory':
 
1560
                    parent_ies[(dirname + '/' + name).strip('/')] = inv_entry
 
1561
                elif kind == 'symlink':
 
1562
                    inv_entry.executable = False
 
1563
                    inv_entry.text_size = None
 
1564
                    inv_entry.symlink_target = utf8_decode(fingerprint)[0]
 
1565
                elif kind == 'tree-reference':
 
1566
                    inv_entry.reference_revision = fingerprint or None
 
1567
                else:
 
1568
                    raise AssertionError("cannot convert entry %r into an InventoryEntry"
 
1569
                            % entry)
 
1570
                # These checks cost us around 40ms on a 55k entry tree
 
1571
                if file_id in inv_byid:
 
1572
                    raise AssertionError('file_id %s already in'
 
1573
                        ' inventory as %s' % (file_id, inv_byid[file_id]))
 
1574
                if name_unicode in parent_ie.children:
 
1575
                    raise AssertionError('name %r already in parent'
 
1576
                        % (name_unicode,))
 
1577
                inv_byid[file_id] = inv_entry
 
1578
                parent_ie.children[name_unicode] = inv_entry
 
1579
        self._inventory = inv
 
1580
 
 
1581
    def get_file_mtime(self, file_id, path=None):
 
1582
        """Return the modification time for this record.
 
1583
 
 
1584
        We return the timestamp of the last-changed revision.
 
1585
        """
 
1586
        # Make sure the file exists
 
1587
        entry = self._get_entry(file_id, path=path)
 
1588
        if entry == (None, None): # do we raise?
 
1589
            return None
 
1590
        parent_index = self._get_parent_index()
 
1591
        last_changed_revision = entry[1][parent_index][4]
 
1592
        return self._repository.get_revision(last_changed_revision).timestamp
 
1593
 
 
1594
    def get_file_sha1(self, file_id, path=None, stat_value=None):
 
1595
        entry = self._get_entry(file_id=file_id, path=path)
 
1596
        parent_index = self._get_parent_index()
 
1597
        parent_details = entry[1][parent_index]
 
1598
        if parent_details[0] == 'f':
 
1599
            return parent_details[1]
 
1600
        return None
 
1601
 
 
1602
    def get_file(self, file_id, path=None):
 
1603
        return StringIO(self.get_file_text(file_id))
 
1604
 
 
1605
    def get_file_lines(self, file_id):
 
1606
        return osutils.split_lines(self.get_file_text(file_id))
 
1607
 
 
1608
    def get_file_size(self, file_id):
 
1609
        """See Tree.get_file_size"""
 
1610
        return self.inventory[file_id].text_size
 
1611
 
 
1612
    def get_file_text(self, file_id):
 
1613
        return list(self.iter_files_bytes([(file_id, None)]))[0][1]
 
1614
 
 
1615
    def get_reference_revision(self, file_id, path=None):
 
1616
        return self.inventory[file_id].reference_revision
 
1617
 
 
1618
    def iter_files_bytes(self, desired_files):
 
1619
        """See Tree.iter_files_bytes.
 
1620
 
 
1621
        This version is implemented on top of Repository.iter_files_bytes"""
 
1622
        parent_index = self._get_parent_index()
 
1623
        repo_desired_files = []
 
1624
        for file_id, identifier in desired_files:
 
1625
            entry = self._get_entry(file_id)
 
1626
            if entry == (None, None):
 
1627
                raise errors.NoSuchId(self, file_id)
 
1628
            repo_desired_files.append((file_id, entry[1][parent_index][4],
 
1629
                                       identifier))
 
1630
        return self._repository.iter_files_bytes(repo_desired_files)
 
1631
 
 
1632
    def get_symlink_target(self, file_id):
 
1633
        entry = self._get_entry(file_id=file_id)
 
1634
        parent_index = self._get_parent_index()
 
1635
        if entry[1][parent_index][0] != 'l':
 
1636
            return None
 
1637
        else:
 
1638
            # At present, none of the tree implementations supports non-ascii
 
1639
            # symlink targets. So we will just assume that the dirstate path is
 
1640
            # correct.
 
1641
            return entry[1][parent_index][1]
 
1642
 
 
1643
    def get_revision_id(self):
 
1644
        """Return the revision id for this tree."""
 
1645
        return self._revision_id
 
1646
 
 
1647
    def _get_inventory(self):
 
1648
        if self._inventory is not None:
 
1649
            return self._inventory
 
1650
        self._must_be_locked()
 
1651
        self._generate_inventory()
 
1652
        return self._inventory
 
1653
 
 
1654
    inventory = property(_get_inventory,
 
1655
                         doc="Inventory of this Tree")
 
1656
 
 
1657
    def get_parent_ids(self):
 
1658
        """The parents of a tree in the dirstate are not cached."""
 
1659
        return self._repository.get_revision(self._revision_id).parent_ids
 
1660
 
 
1661
    def has_filename(self, filename):
 
1662
        return bool(self.path2id(filename))
 
1663
 
 
1664
    def kind(self, file_id):
 
1665
        entry = self._get_entry(file_id=file_id)[1]
 
1666
        if entry is None:
 
1667
            raise errors.NoSuchId(tree=self, file_id=file_id)
 
1668
        return dirstate.DirState._minikind_to_kind[entry[1][0]]
 
1669
 
 
1670
    def stored_kind(self, file_id):
 
1671
        """See Tree.stored_kind"""
 
1672
        return self.kind(file_id)
 
1673
 
 
1674
    def path_content_summary(self, path):
 
1675
        """See Tree.path_content_summary."""
 
1676
        id = self.inventory.path2id(path)
 
1677
        if id is None:
 
1678
            return ('missing', None, None, None)
 
1679
        entry = self._inventory[id]
 
1680
        kind = entry.kind
 
1681
        if kind == 'file':
 
1682
            return (kind, entry.text_size, entry.executable, entry.text_sha1)
 
1683
        elif kind == 'symlink':
 
1684
            return (kind, None, None, entry.symlink_target)
 
1685
        else:
 
1686
            return (kind, None, None, None)
 
1687
 
 
1688
    def is_executable(self, file_id, path=None):
 
1689
        ie = self.inventory[file_id]
 
1690
        if ie.kind != "file":
 
1691
            return None
 
1692
        return ie.executable
 
1693
 
 
1694
    def list_files(self, include_root=False):
 
1695
        # We use a standard implementation, because DirStateRevisionTree is
 
1696
        # dealing with one of the parents of the current state
 
1697
        inv = self._get_inventory()
 
1698
        entries = inv.iter_entries()
 
1699
        if self.inventory.root is not None and not include_root:
 
1700
            entries.next()
 
1701
        for path, entry in entries:
 
1702
            yield path, 'V', entry.kind, entry.file_id, entry
 
1703
 
 
1704
    def lock_read(self):
 
1705
        """Lock the tree for a set of operations."""
 
1706
        if not self._locked:
 
1707
            self._repository.lock_read()
 
1708
            if self._dirstate._lock_token is None:
 
1709
                self._dirstate.lock_read()
 
1710
                self._dirstate_locked = True
 
1711
        self._locked += 1
 
1712
 
 
1713
    def _must_be_locked(self):
 
1714
        if not self._locked:
 
1715
            raise errors.ObjectNotLocked(self)
 
1716
 
 
1717
    @needs_read_lock
 
1718
    def path2id(self, path):
 
1719
        """Return the id for path in this tree."""
 
1720
        # lookup by path: faster than splitting and walking the ivnentory.
 
1721
        entry = self._get_entry(path=path)
 
1722
        if entry == (None, None):
 
1723
            return None
 
1724
        return entry[0][2]
 
1725
 
 
1726
    def unlock(self):
 
1727
        """Unlock, freeing any cache memory used during the lock."""
 
1728
        # outside of a lock, the inventory is suspect: release it.
 
1729
        self._locked -=1
 
1730
        if not self._locked:
 
1731
            self._inventory = None
 
1732
            self._locked = 0
 
1733
            if self._dirstate_locked:
 
1734
                self._dirstate.unlock()
 
1735
                self._dirstate_locked = False
 
1736
            self._repository.unlock()
 
1737
 
 
1738
    @needs_read_lock
 
1739
    def supports_tree_reference(self):
 
1740
        return self._repo_supports_tree_reference
 
1741
 
 
1742
    def walkdirs(self, prefix=""):
 
1743
        # TODO: jam 20070215 This is the lazy way by using the RevisionTree
 
1744
        # implementation based on an inventory.  
 
1745
        # This should be cleaned up to use the much faster Dirstate code
 
1746
        # So for now, we just build up the parent inventory, and extract
 
1747
        # it the same way RevisionTree does.
 
1748
        _directory = 'directory'
 
1749
        inv = self._get_inventory()
 
1750
        top_id = inv.path2id(prefix)
 
1751
        if top_id is None:
 
1752
            pending = []
 
1753
        else:
 
1754
            pending = [(prefix, top_id)]
 
1755
        while pending:
 
1756
            dirblock = []
 
1757
            relpath, file_id = pending.pop()
 
1758
            # 0 - relpath, 1- file-id
 
1759
            if relpath:
 
1760
                relroot = relpath + '/'
 
1761
            else:
 
1762
                relroot = ""
 
1763
            # FIXME: stash the node in pending
 
1764
            entry = inv[file_id]
 
1765
            for name, child in entry.sorted_children():
 
1766
                toppath = relroot + name
 
1767
                dirblock.append((toppath, name, child.kind, None,
 
1768
                    child.file_id, child.kind
 
1769
                    ))
 
1770
            yield (relpath, entry.file_id), dirblock
 
1771
            # push the user specified dirs from dirblock
 
1772
            for dir in reversed(dirblock):
 
1773
                if dir[2] == _directory:
 
1774
                    pending.append((dir[0], dir[4]))
 
1775
 
 
1776
 
 
1777
class InterDirStateTree(InterTree):
 
1778
    """Fast path optimiser for changes_from with dirstate trees.
 
1779
    
 
1780
    This is used only when both trees are in the dirstate working file, and 
 
1781
    the source is any parent within the dirstate, and the destination is 
 
1782
    the current working tree of the same dirstate.
 
1783
    """
 
1784
    # this could be generalized to allow comparisons between any trees in the
 
1785
    # dirstate, and possibly between trees stored in different dirstates.
 
1786
 
 
1787
    def __init__(self, source, target):
 
1788
        super(InterDirStateTree, self).__init__(source, target)
 
1789
        if not InterDirStateTree.is_compatible(source, target):
 
1790
            raise Exception, "invalid source %r and target %r" % (source, target)
 
1791
 
 
1792
    @staticmethod
 
1793
    def make_source_parent_tree(source, target):
 
1794
        """Change the source tree into a parent of the target."""
 
1795
        revid = source.commit('record tree')
 
1796
        target.branch.repository.fetch(source.branch.repository, revid)
 
1797
        target.set_parent_ids([revid])
 
1798
        return target.basis_tree(), target
 
1799
 
 
1800
    _matching_from_tree_format = WorkingTreeFormat4()
 
1801
    _matching_to_tree_format = WorkingTreeFormat4()
 
1802
    _test_mutable_trees_to_test_trees = make_source_parent_tree
 
1803
 
 
1804
    def iter_changes(self, include_unchanged=False,
 
1805
                      specific_files=None, pb=None, extra_trees=[],
 
1806
                      require_versioned=True, want_unversioned=False):
 
1807
        """Return the changes from source to target.
 
1808
 
 
1809
        :return: An iterator that yields tuples. See InterTree.iter_changes
 
1810
            for details.
 
1811
        :param specific_files: An optional list of file paths to restrict the
 
1812
            comparison to. When mapping filenames to ids, all matches in all
 
1813
            trees (including optional extra_trees) are used, and all children of
 
1814
            matched directories are included.
 
1815
        :param include_unchanged: An optional boolean requesting the inclusion of
 
1816
            unchanged entries in the result.
 
1817
        :param extra_trees: An optional list of additional trees to use when
 
1818
            mapping the contents of specific_files (paths) to file_ids.
 
1819
        :param require_versioned: If True, all files in specific_files must be
 
1820
            versioned in one of source, target, extra_trees or
 
1821
            PathsNotVersionedError is raised.
 
1822
        :param want_unversioned: Should unversioned files be returned in the
 
1823
            output. An unversioned file is defined as one with (False, False)
 
1824
            for the versioned pair.
 
1825
        """
 
1826
        utf8_decode = cache_utf8._utf8_decode
 
1827
        _minikind_to_kind = dirstate.DirState._minikind_to_kind
 
1828
        cmp_by_dirs = dirstate.cmp_by_dirs
 
1829
        # NB: show_status depends on being able to pass in non-versioned files
 
1830
        # and report them as unknown
 
1831
        # TODO: handle extra trees in the dirstate.
 
1832
        if (extra_trees or specific_files == []):
 
1833
            # we can't fast-path these cases (yet)
 
1834
            for f in super(InterDirStateTree, self).iter_changes(
 
1835
                include_unchanged, specific_files, pb, extra_trees,
 
1836
                require_versioned, want_unversioned=want_unversioned):
 
1837
                yield f
 
1838
            return
 
1839
        parent_ids = self.target.get_parent_ids()
 
1840
        if not (self.source._revision_id in parent_ids
 
1841
                or self.source._revision_id == NULL_REVISION):
 
1842
            raise AssertionError(
 
1843
                "revision {%s} is not stored in {%s}, but %s "
 
1844
                "can only be used for trees stored in the dirstate"
 
1845
                % (self.source._revision_id, self.target, self.iter_changes))
 
1846
        target_index = 0
 
1847
        if self.source._revision_id == NULL_REVISION:
 
1848
            source_index = None
 
1849
            indices = (target_index,)
 
1850
        else:
 
1851
            if not (self.source._revision_id in parent_ids):
 
1852
                raise AssertionError(
 
1853
                    "Failure: source._revision_id: %s not in target.parent_ids(%s)" % (
 
1854
                    self.source._revision_id, parent_ids))
 
1855
            source_index = 1 + parent_ids.index(self.source._revision_id)
 
1856
            indices = (source_index, target_index)
 
1857
        # -- make all specific_files utf8 --
 
1858
        if specific_files:
 
1859
            specific_files_utf8 = set()
 
1860
            for path in specific_files:
 
1861
                specific_files_utf8.add(path.encode('utf8'))
 
1862
            specific_files = specific_files_utf8
 
1863
        else:
 
1864
            specific_files = set([''])
 
1865
        # -- specific_files is now a utf8 path set --
 
1866
        # -- get the state object and prepare it.
 
1867
        state = self.target.current_dirstate()
 
1868
        state._read_dirblocks_if_needed()
 
1869
        def _entries_for_path(path):
 
1870
            """Return a list with all the entries that match path for all ids.
 
1871
            """
 
1872
            dirname, basename = os.path.split(path)
 
1873
            key = (dirname, basename, '')
 
1874
            block_index, present = state._find_block_index_from_key(key)
 
1875
            if not present:
 
1876
                # the block which should contain path is absent.
 
1877
                return []
 
1878
            result = []
 
1879
            block = state._dirblocks[block_index][1]
 
1880
            entry_index, _ = state._find_entry_index(key, block)
 
1881
            # we may need to look at multiple entries at this path: walk while the specific_files match.
 
1882
            while (entry_index < len(block) and
 
1883
                block[entry_index][0][0:2] == key[0:2]):
 
1884
                result.append(block[entry_index])
 
1885
                entry_index += 1
 
1886
            return result
 
1887
        if require_versioned:
 
1888
            # -- check all supplied paths are versioned in a search tree. --
 
1889
            all_versioned = True
 
1890
            for path in specific_files:
 
1891
                path_entries = _entries_for_path(path)
 
1892
                if not path_entries:
 
1893
                    # this specified path is not present at all: error
 
1894
                    all_versioned = False
 
1895
                    break
 
1896
                found_versioned = False
 
1897
                # for each id at this path
 
1898
                for entry in path_entries:
 
1899
                    # for each tree.
 
1900
                    for index in indices:
 
1901
                        if entry[1][index][0] != 'a': # absent
 
1902
                            found_versioned = True
 
1903
                            # all good: found a versioned cell
 
1904
                            break
 
1905
                if not found_versioned:
 
1906
                    # none of the indexes was not 'absent' at all ids for this
 
1907
                    # path.
 
1908
                    all_versioned = False
 
1909
                    break
 
1910
            if not all_versioned:
 
1911
                raise errors.PathsNotVersionedError(specific_files)
 
1912
        # -- remove redundancy in supplied specific_files to prevent over-scanning --
 
1913
        search_specific_files = set()
 
1914
        for path in specific_files:
 
1915
            other_specific_files = specific_files.difference(set([path]))
 
1916
            if not osutils.is_inside_any(other_specific_files, path):
 
1917
                # this is a top level path, we must check it.
 
1918
                search_specific_files.add(path)
 
1919
        # sketch: 
 
1920
        # compare source_index and target_index at or under each element of search_specific_files.
 
1921
        # follow the following comparison table. Note that we only want to do diff operations when
 
1922
        # the target is fdl because thats when the walkdirs logic will have exposed the pathinfo 
 
1923
        # for the target.
 
1924
        # cases:
 
1925
        # 
 
1926
        # Source | Target | disk | action
 
1927
        #   r    | fdlt   |      | add source to search, add id path move and perform
 
1928
        #        |        |      | diff check on source-target
 
1929
        #   r    | fdlt   |  a   | dangling file that was present in the basis. 
 
1930
        #        |        |      | ???
 
1931
        #   r    |  a     |      | add source to search
 
1932
        #   r    |  a     |  a   | 
 
1933
        #   r    |  r     |      | this path is present in a non-examined tree, skip.
 
1934
        #   r    |  r     |  a   | this path is present in a non-examined tree, skip.
 
1935
        #   a    | fdlt   |      | add new id
 
1936
        #   a    | fdlt   |  a   | dangling locally added file, skip
 
1937
        #   a    |  a     |      | not present in either tree, skip
 
1938
        #   a    |  a     |  a   | not present in any tree, skip
 
1939
        #   a    |  r     |      | not present in either tree at this path, skip as it
 
1940
        #        |        |      | may not be selected by the users list of paths.
 
1941
        #   a    |  r     |  a   | not present in either tree at this path, skip as it
 
1942
        #        |        |      | may not be selected by the users list of paths.
 
1943
        #  fdlt  | fdlt   |      | content in both: diff them
 
1944
        #  fdlt  | fdlt   |  a   | deleted locally, but not unversioned - show as deleted ?
 
1945
        #  fdlt  |  a     |      | unversioned: output deleted id for now
 
1946
        #  fdlt  |  a     |  a   | unversioned and deleted: output deleted id
 
1947
        #  fdlt  |  r     |      | relocated in this tree, so add target to search.
 
1948
        #        |        |      | Dont diff, we will see an r,fd; pair when we reach
 
1949
        #        |        |      | this id at the other path.
 
1950
        #  fdlt  |  r     |  a   | relocated in this tree, so add target to search.
 
1951
        #        |        |      | Dont diff, we will see an r,fd; pair when we reach
 
1952
        #        |        |      | this id at the other path.
 
1953
 
 
1954
        # for all search_indexs in each path at or under each element of
 
1955
        # search_specific_files, if the detail is relocated: add the id, and add the
 
1956
        # relocated path as one to search if its not searched already. If the
 
1957
        # detail is not relocated, add the id.
 
1958
        searched_specific_files = set()
 
1959
        NULL_PARENT_DETAILS = dirstate.DirState.NULL_PARENT_DETAILS
 
1960
        # Using a list so that we can access the values and change them in
 
1961
        # nested scope. Each one is [path, file_id, entry]
 
1962
        last_source_parent = [None, None]
 
1963
        last_target_parent = [None, None]
 
1964
 
 
1965
        use_filesystem_for_exec = (sys.platform != 'win32')
 
1966
 
 
1967
        # Just a sentry, so that _process_entry can say that this
 
1968
        # record is handled, but isn't interesting to process (unchanged)
 
1969
        uninteresting = object()
 
1970
 
 
1971
 
 
1972
        old_dirname_to_file_id = {}
 
1973
        new_dirname_to_file_id = {}
 
1974
        # TODO: jam 20070516 - Avoid the _get_entry lookup overhead by
 
1975
        #       keeping a cache of directories that we have seen.
 
1976
 
 
1977
        def _process_entry(entry, path_info):
 
1978
            """Compare an entry and real disk to generate delta information.
 
1979
 
 
1980
            :param path_info: top_relpath, basename, kind, lstat, abspath for
 
1981
                the path of entry. If None, then the path is considered absent.
 
1982
                (Perhaps we should pass in a concrete entry for this ?)
 
1983
                Basename is returned as a utf8 string because we expect this
 
1984
                tuple will be ignored, and don't want to take the time to
 
1985
                decode.
 
1986
            :return: None if these don't match
 
1987
                     A tuple of information about the change, or
 
1988
                     the object 'uninteresting' if these match, but are
 
1989
                     basically identical.
 
1990
            """
 
1991
            if source_index is None:
 
1992
                source_details = NULL_PARENT_DETAILS
 
1993
            else:
 
1994
                source_details = entry[1][source_index]
 
1995
            target_details = entry[1][target_index]
 
1996
            target_minikind = target_details[0]
 
1997
            if path_info is not None and target_minikind in 'fdlt':
 
1998
                if not (target_index == 0):
 
1999
                    raise AssertionError()
 
2000
                link_or_sha1 = state.update_entry(entry, abspath=path_info[4],
 
2001
                                                  stat_value=path_info[3])
 
2002
                # The entry may have been modified by update_entry
 
2003
                target_details = entry[1][target_index]
 
2004
                target_minikind = target_details[0]
 
2005
            else:
 
2006
                link_or_sha1 = None
 
2007
            file_id = entry[0][2]
 
2008
            source_minikind = source_details[0]
 
2009
            if source_minikind in 'fdltr' and target_minikind in 'fdlt':
 
2010
                # claimed content in both: diff
 
2011
                #   r    | fdlt   |      | add source to search, add id path move and perform
 
2012
                #        |        |      | diff check on source-target
 
2013
                #   r    | fdlt   |  a   | dangling file that was present in the basis.
 
2014
                #        |        |      | ???
 
2015
                if source_minikind in 'r':
 
2016
                    # add the source to the search path to find any children it
 
2017
                    # has.  TODO ? : only add if it is a container ?
 
2018
                    if not osutils.is_inside_any(searched_specific_files,
 
2019
                                                 source_details[1]):
 
2020
                        search_specific_files.add(source_details[1])
 
2021
                    # generate the old path; this is needed for stating later
 
2022
                    # as well.
 
2023
                    old_path = source_details[1]
 
2024
                    old_dirname, old_basename = os.path.split(old_path)
 
2025
                    path = pathjoin(entry[0][0], entry[0][1])
 
2026
                    old_entry = state._get_entry(source_index,
 
2027
                                                 path_utf8=old_path)
 
2028
                    # update the source details variable to be the real
 
2029
                    # location.
 
2030
                    if old_entry == (None, None):
 
2031
                        raise errors.CorruptDirstate(state._filename,
 
2032
                            "entry '%s/%s' is considered renamed from %r"
 
2033
                            " but source does not exist\n"
 
2034
                            "entry: %s" % (entry[0][0], entry[0][1], old_path, entry))
 
2035
                    source_details = old_entry[1][source_index]
 
2036
                    source_minikind = source_details[0]
 
2037
                else:
 
2038
                    old_dirname = entry[0][0]
 
2039
                    old_basename = entry[0][1]
 
2040
                    old_path = path = None
 
2041
                if path_info is None:
 
2042
                    # the file is missing on disk, show as removed.
 
2043
                    content_change = True
 
2044
                    target_kind = None
 
2045
                    target_exec = False
 
2046
                else:
 
2047
                    # source and target are both versioned and disk file is present.
 
2048
                    target_kind = path_info[2]
 
2049
                    if target_kind == 'directory':
 
2050
                        if path is None:
 
2051
                            old_path = path = pathjoin(old_dirname, old_basename)
 
2052
                        new_dirname_to_file_id[path] = file_id
 
2053
                        if source_minikind != 'd':
 
2054
                            content_change = True
 
2055
                        else:
 
2056
                            # directories have no fingerprint
 
2057
                            content_change = False
 
2058
                        target_exec = False
 
2059
                    elif target_kind == 'file':
 
2060
                        if source_minikind != 'f':
 
2061
                            content_change = True
 
2062
                        else:
 
2063
                            # We could check the size, but we already have the
 
2064
                            # sha1 hash.
 
2065
                            content_change = (link_or_sha1 != source_details[1])
 
2066
                        # Target details is updated at update_entry time
 
2067
                        if use_filesystem_for_exec:
 
2068
                            # We don't need S_ISREG here, because we are sure
 
2069
                            # we are dealing with a file.
 
2070
                            target_exec = bool(stat.S_IEXEC & path_info[3].st_mode)
 
2071
                        else:
 
2072
                            target_exec = target_details[3]
 
2073
                    elif target_kind == 'symlink':
 
2074
                        if source_minikind != 'l':
 
2075
                            content_change = True
 
2076
                        else:
 
2077
                            content_change = (link_or_sha1 != source_details[1])
 
2078
                        target_exec = False
 
2079
                    elif target_kind == 'tree-reference':
 
2080
                        if source_minikind != 't':
 
2081
                            content_change = True
 
2082
                        else:
 
2083
                            content_change = False
 
2084
                        target_exec = False
 
2085
                    else:
 
2086
                        raise Exception, "unknown kind %s" % path_info[2]
 
2087
                if source_minikind == 'd':
 
2088
                    if path is None:
 
2089
                        old_path = path = pathjoin(old_dirname, old_basename)
 
2090
                    old_dirname_to_file_id[old_path] = file_id
 
2091
                # parent id is the entry for the path in the target tree
 
2092
                if old_dirname == last_source_parent[0]:
 
2093
                    source_parent_id = last_source_parent[1]
 
2094
                else:
 
2095
                    try:
 
2096
                        source_parent_id = old_dirname_to_file_id[old_dirname]
 
2097
                    except KeyError:
 
2098
                        source_parent_entry = state._get_entry(source_index,
 
2099
                                                               path_utf8=old_dirname)
 
2100
                        source_parent_id = source_parent_entry[0][2]
 
2101
                    if source_parent_id == entry[0][2]:
 
2102
                        # This is the root, so the parent is None
 
2103
                        source_parent_id = None
 
2104
                    else:
 
2105
                        last_source_parent[0] = old_dirname
 
2106
                        last_source_parent[1] = source_parent_id
 
2107
                new_dirname = entry[0][0]
 
2108
                if new_dirname == last_target_parent[0]:
 
2109
                    target_parent_id = last_target_parent[1]
 
2110
                else:
 
2111
                    try:
 
2112
                        target_parent_id = new_dirname_to_file_id[new_dirname]
 
2113
                    except KeyError:
 
2114
                        # TODO: We don't always need to do the lookup, because the
 
2115
                        #       parent entry will be the same as the source entry.
 
2116
                        target_parent_entry = state._get_entry(target_index,
 
2117
                                                               path_utf8=new_dirname)
 
2118
                        if target_parent_entry == (None, None):
 
2119
                            raise AssertionError(
 
2120
                                "Could not find target parent in wt: %s\nparent of: %s"
 
2121
                                % (new_dirname, entry))
 
2122
                        target_parent_id = target_parent_entry[0][2]
 
2123
                    if target_parent_id == entry[0][2]:
 
2124
                        # This is the root, so the parent is None
 
2125
                        target_parent_id = None
 
2126
                    else:
 
2127
                        last_target_parent[0] = new_dirname
 
2128
                        last_target_parent[1] = target_parent_id
 
2129
 
 
2130
                source_exec = source_details[3]
 
2131
                if (include_unchanged
 
2132
                    or content_change
 
2133
                    or source_parent_id != target_parent_id
 
2134
                    or old_basename != entry[0][1]
 
2135
                    or source_exec != target_exec
 
2136
                    ):
 
2137
                    if old_path is None:
 
2138
                        old_path = path = pathjoin(old_dirname, old_basename)
 
2139
                        old_path_u = utf8_decode(old_path)[0]
 
2140
                        path_u = old_path_u
 
2141
                    else:
 
2142
                        old_path_u = utf8_decode(old_path)[0]
 
2143
                        if old_path == path:
 
2144
                            path_u = old_path_u
 
2145
                        else:
 
2146
                            path_u = utf8_decode(path)[0]
 
2147
                    source_kind = _minikind_to_kind[source_minikind]
 
2148
                    return (entry[0][2],
 
2149
                           (old_path_u, path_u),
 
2150
                           content_change,
 
2151
                           (True, True),
 
2152
                           (source_parent_id, target_parent_id),
 
2153
                           (utf8_decode(old_basename)[0], utf8_decode(entry[0][1])[0]),
 
2154
                           (source_kind, target_kind),
 
2155
                           (source_exec, target_exec))
 
2156
                else:
 
2157
                    return uninteresting
 
2158
            elif source_minikind in 'a' and target_minikind in 'fdlt':
 
2159
                # looks like a new file
 
2160
                if path_info is not None:
 
2161
                    path = pathjoin(entry[0][0], entry[0][1])
 
2162
                    # parent id is the entry for the path in the target tree
 
2163
                    # TODO: these are the same for an entire directory: cache em.
 
2164
                    parent_id = state._get_entry(target_index,
 
2165
                                                 path_utf8=entry[0][0])[0][2]
 
2166
                    if parent_id == entry[0][2]:
 
2167
                        parent_id = None
 
2168
                    if use_filesystem_for_exec:
 
2169
                        # We need S_ISREG here, because we aren't sure if this
 
2170
                        # is a file or not.
 
2171
                        target_exec = bool(
 
2172
                            stat.S_ISREG(path_info[3].st_mode)
 
2173
                            and stat.S_IEXEC & path_info[3].st_mode)
 
2174
                    else:
 
2175
                        target_exec = target_details[3]
 
2176
                    return (entry[0][2],
 
2177
                           (None, utf8_decode(path)[0]),
 
2178
                           True,
 
2179
                           (False, True),
 
2180
                           (None, parent_id),
 
2181
                           (None, utf8_decode(entry[0][1])[0]),
 
2182
                           (None, path_info[2]),
 
2183
                           (None, target_exec))
 
2184
                else:
 
2185
                    # but its not on disk: we deliberately treat this as just
 
2186
                    # never-present. (Why ?! - RBC 20070224)
 
2187
                    pass
 
2188
            elif source_minikind in 'fdlt' and target_minikind in 'a':
 
2189
                # unversioned, possibly, or possibly not deleted: we dont care.
 
2190
                # if its still on disk, *and* theres no other entry at this
 
2191
                # path [we dont know this in this routine at the moment -
 
2192
                # perhaps we should change this - then it would be an unknown.
 
2193
                old_path = pathjoin(entry[0][0], entry[0][1])
 
2194
                # parent id is the entry for the path in the target tree
 
2195
                parent_id = state._get_entry(source_index, path_utf8=entry[0][0])[0][2]
 
2196
                if parent_id == entry[0][2]:
 
2197
                    parent_id = None
 
2198
                return (entry[0][2],
 
2199
                       (utf8_decode(old_path)[0], None),
 
2200
                       True,
 
2201
                       (True, False),
 
2202
                       (parent_id, None),
 
2203
                       (utf8_decode(entry[0][1])[0], None),
 
2204
                       (_minikind_to_kind[source_minikind], None),
 
2205
                       (source_details[3], None))
 
2206
            elif source_minikind in 'fdlt' and target_minikind in 'r':
 
2207
                # a rename; could be a true rename, or a rename inherited from
 
2208
                # a renamed parent. TODO: handle this efficiently. Its not
 
2209
                # common case to rename dirs though, so a correct but slow
 
2210
                # implementation will do.
 
2211
                if not osutils.is_inside_any(searched_specific_files, target_details[1]):
 
2212
                    search_specific_files.add(target_details[1])
 
2213
            elif source_minikind in 'ra' and target_minikind in 'ra':
 
2214
                # neither of the selected trees contain this file,
 
2215
                # so skip over it. This is not currently directly tested, but
 
2216
                # is indirectly via test_too_much.TestCommands.test_conflicts.
 
2217
                pass
 
2218
            else:
 
2219
                raise AssertionError("don't know how to compare "
 
2220
                    "source_minikind=%r, target_minikind=%r"
 
2221
                    % (source_minikind, target_minikind))
 
2222
                ## import pdb;pdb.set_trace()
 
2223
            return None
 
2224
 
 
2225
        while search_specific_files:
 
2226
            # TODO: the pending list should be lexically sorted?  the
 
2227
            # interface doesn't require it.
 
2228
            current_root = search_specific_files.pop()
 
2229
            current_root_unicode = current_root.decode('utf8')
 
2230
            searched_specific_files.add(current_root)
 
2231
            # process the entries for this containing directory: the rest will be
 
2232
            # found by their parents recursively.
 
2233
            root_entries = _entries_for_path(current_root)
 
2234
            root_abspath = self.target.abspath(current_root_unicode)
 
2235
            try:
 
2236
                root_stat = os.lstat(root_abspath)
 
2237
            except OSError, e:
 
2238
                if e.errno == errno.ENOENT:
 
2239
                    # the path does not exist: let _process_entry know that.
 
2240
                    root_dir_info = None
 
2241
                else:
 
2242
                    # some other random error: hand it up.
 
2243
                    raise
 
2244
            else:
 
2245
                root_dir_info = ('', current_root,
 
2246
                    osutils.file_kind_from_stat_mode(root_stat.st_mode), root_stat,
 
2247
                    root_abspath)
 
2248
                if root_dir_info[2] == 'directory':
 
2249
                    if self.target._directory_is_tree_reference(
 
2250
                        current_root.decode('utf8')):
 
2251
                        root_dir_info = root_dir_info[:2] + \
 
2252
                            ('tree-reference',) + root_dir_info[3:]
 
2253
 
 
2254
            if not root_entries and not root_dir_info:
 
2255
                # this specified path is not present at all, skip it.
 
2256
                continue
 
2257
            path_handled = False
 
2258
            for entry in root_entries:
 
2259
                result = _process_entry(entry, root_dir_info)
 
2260
                if result is not None:
 
2261
                    path_handled = True
 
2262
                    if result is not uninteresting:
 
2263
                        yield result
 
2264
            if want_unversioned and not path_handled and root_dir_info:
 
2265
                new_executable = bool(
 
2266
                    stat.S_ISREG(root_dir_info[3].st_mode)
 
2267
                    and stat.S_IEXEC & root_dir_info[3].st_mode)
 
2268
                yield (None,
 
2269
                       (None, current_root_unicode),
 
2270
                       True,
 
2271
                       (False, False),
 
2272
                       (None, None),
 
2273
                       (None, splitpath(current_root_unicode)[-1]),
 
2274
                       (None, root_dir_info[2]),
 
2275
                       (None, new_executable)
 
2276
                      )
 
2277
            initial_key = (current_root, '', '')
 
2278
            block_index, _ = state._find_block_index_from_key(initial_key)
 
2279
            if block_index == 0:
 
2280
                # we have processed the total root already, but because the
 
2281
                # initial key matched it we should skip it here.
 
2282
                block_index +=1
 
2283
            if root_dir_info and root_dir_info[2] == 'tree-reference':
 
2284
                current_dir_info = None
 
2285
            else:
 
2286
                dir_iterator = osutils._walkdirs_utf8(root_abspath, prefix=current_root)
 
2287
                try:
 
2288
                    current_dir_info = dir_iterator.next()
 
2289
                except OSError, e:
 
2290
                    # on win32, python2.4 has e.errno == ERROR_DIRECTORY, but
 
2291
                    # python 2.5 has e.errno == EINVAL,
 
2292
                    #            and e.winerror == ERROR_DIRECTORY
 
2293
                    e_winerror = getattr(e, 'winerror', None)
 
2294
                    win_errors = (ERROR_DIRECTORY, ERROR_PATH_NOT_FOUND)
 
2295
                    # there may be directories in the inventory even though
 
2296
                    # this path is not a file on disk: so mark it as end of
 
2297
                    # iterator
 
2298
                    if e.errno in (errno.ENOENT, errno.ENOTDIR, errno.EINVAL):
 
2299
                        current_dir_info = None
 
2300
                    elif (sys.platform == 'win32'
 
2301
                          and (e.errno in win_errors
 
2302
                               or e_winerror in win_errors)):
 
2303
                        current_dir_info = None
 
2304
                    else:
 
2305
                        raise
 
2306
                else:
 
2307
                    if current_dir_info[0][0] == '':
 
2308
                        # remove .bzr from iteration
 
2309
                        bzr_index = bisect_left(current_dir_info[1], ('.bzr',))
 
2310
                        if current_dir_info[1][bzr_index][0] != '.bzr':
 
2311
                            raise AssertionError()
 
2312
                        del current_dir_info[1][bzr_index]
 
2313
            # walk until both the directory listing and the versioned metadata
 
2314
            # are exhausted. 
 
2315
            if (block_index < len(state._dirblocks) and
 
2316
                osutils.is_inside(current_root, state._dirblocks[block_index][0])):
 
2317
                current_block = state._dirblocks[block_index]
 
2318
            else:
 
2319
                current_block = None
 
2320
            while (current_dir_info is not None or
 
2321
                   current_block is not None):
 
2322
                if (current_dir_info and current_block
 
2323
                    and current_dir_info[0][0] != current_block[0]):
 
2324
                    if cmp_by_dirs(current_dir_info[0][0], current_block[0]) < 0:
 
2325
                        # filesystem data refers to paths not covered by the dirblock.
 
2326
                        # this has two possibilities:
 
2327
                        # A) it is versioned but empty, so there is no block for it
 
2328
                        # B) it is not versioned.
 
2329
 
 
2330
                        # if (A) then we need to recurse into it to check for
 
2331
                        # new unknown files or directories.
 
2332
                        # if (B) then we should ignore it, because we don't
 
2333
                        # recurse into unknown directories.
 
2334
                        path_index = 0
 
2335
                        while path_index < len(current_dir_info[1]):
 
2336
                                current_path_info = current_dir_info[1][path_index]
 
2337
                                if want_unversioned:
 
2338
                                    if current_path_info[2] == 'directory':
 
2339
                                        if self.target._directory_is_tree_reference(
 
2340
                                            current_path_info[0].decode('utf8')):
 
2341
                                            current_path_info = current_path_info[:2] + \
 
2342
                                                ('tree-reference',) + current_path_info[3:]
 
2343
                                    new_executable = bool(
 
2344
                                        stat.S_ISREG(current_path_info[3].st_mode)
 
2345
                                        and stat.S_IEXEC & current_path_info[3].st_mode)
 
2346
                                    yield (None,
 
2347
                                        (None, utf8_decode(current_path_info[0])[0]),
 
2348
                                        True,
 
2349
                                        (False, False),
 
2350
                                        (None, None),
 
2351
                                        (None, utf8_decode(current_path_info[1])[0]),
 
2352
                                        (None, current_path_info[2]),
 
2353
                                        (None, new_executable))
 
2354
                                # dont descend into this unversioned path if it is
 
2355
                                # a dir
 
2356
                                if current_path_info[2] in ('directory',
 
2357
                                                            'tree-reference'):
 
2358
                                    del current_dir_info[1][path_index]
 
2359
                                    path_index -= 1
 
2360
                                path_index += 1
 
2361
 
 
2362
                        # This dir info has been handled, go to the next
 
2363
                        try:
 
2364
                            current_dir_info = dir_iterator.next()
 
2365
                        except StopIteration:
 
2366
                            current_dir_info = None
 
2367
                    else:
 
2368
                        # We have a dirblock entry for this location, but there
 
2369
                        # is no filesystem path for this. This is most likely
 
2370
                        # because a directory was removed from the disk.
 
2371
                        # We don't have to report the missing directory,
 
2372
                        # because that should have already been handled, but we
 
2373
                        # need to handle all of the files that are contained
 
2374
                        # within.
 
2375
                        for current_entry in current_block[1]:
 
2376
                            # entry referring to file not present on disk.
 
2377
                            # advance the entry only, after processing.
 
2378
                            result = _process_entry(current_entry, None)
 
2379
                            if result is not None:
 
2380
                                if result is not uninteresting:
 
2381
                                    yield result
 
2382
                        block_index +=1
 
2383
                        if (block_index < len(state._dirblocks) and
 
2384
                            osutils.is_inside(current_root,
 
2385
                                              state._dirblocks[block_index][0])):
 
2386
                            current_block = state._dirblocks[block_index]
 
2387
                        else:
 
2388
                            current_block = None
 
2389
                    continue
 
2390
                entry_index = 0
 
2391
                if current_block and entry_index < len(current_block[1]):
 
2392
                    current_entry = current_block[1][entry_index]
 
2393
                else:
 
2394
                    current_entry = None
 
2395
                advance_entry = True
 
2396
                path_index = 0
 
2397
                if current_dir_info and path_index < len(current_dir_info[1]):
 
2398
                    current_path_info = current_dir_info[1][path_index]
 
2399
                    if current_path_info[2] == 'directory':
 
2400
                        if self.target._directory_is_tree_reference(
 
2401
                            current_path_info[0].decode('utf8')):
 
2402
                            current_path_info = current_path_info[:2] + \
 
2403
                                ('tree-reference',) + current_path_info[3:]
 
2404
                else:
 
2405
                    current_path_info = None
 
2406
                advance_path = True
 
2407
                path_handled = False
 
2408
                while (current_entry is not None or
 
2409
                    current_path_info is not None):
 
2410
                    if current_entry is None:
 
2411
                        # the check for path_handled when the path is adnvaced
 
2412
                        # will yield this path if needed.
 
2413
                        pass
 
2414
                    elif current_path_info is None:
 
2415
                        # no path is fine: the per entry code will handle it.
 
2416
                        result = _process_entry(current_entry, current_path_info)
 
2417
                        if result is not None:
 
2418
                            if result is not uninteresting:
 
2419
                                yield result
 
2420
                    elif (current_entry[0][1] != current_path_info[1]
 
2421
                          or current_entry[1][target_index][0] in 'ar'):
 
2422
                        # The current path on disk doesn't match the dirblock
 
2423
                        # record. Either the dirblock is marked as absent, or
 
2424
                        # the file on disk is not present at all in the
 
2425
                        # dirblock. Either way, report about the dirblock
 
2426
                        # entry, and let other code handle the filesystem one.
 
2427
 
 
2428
                        # Compare the basename for these files to determine
 
2429
                        # which comes first
 
2430
                        if current_path_info[1] < current_entry[0][1]:
 
2431
                            # extra file on disk: pass for now, but only
 
2432
                            # increment the path, not the entry
 
2433
                            advance_entry = False
 
2434
                        else:
 
2435
                            # entry referring to file not present on disk.
 
2436
                            # advance the entry only, after processing.
 
2437
                            result = _process_entry(current_entry, None)
 
2438
                            if result is not None:
 
2439
                                if result is not uninteresting:
 
2440
                                    yield result
 
2441
                            advance_path = False
 
2442
                    else:
 
2443
                        result = _process_entry(current_entry, current_path_info)
 
2444
                        if result is not None:
 
2445
                            path_handled = True
 
2446
                            if result is not uninteresting:
 
2447
                                yield result
 
2448
                    if advance_entry and current_entry is not None:
 
2449
                        entry_index += 1
 
2450
                        if entry_index < len(current_block[1]):
 
2451
                            current_entry = current_block[1][entry_index]
 
2452
                        else:
 
2453
                            current_entry = None
 
2454
                    else:
 
2455
                        advance_entry = True # reset the advance flaga
 
2456
                    if advance_path and current_path_info is not None:
 
2457
                        if not path_handled:
 
2458
                            # unversioned in all regards
 
2459
                            if want_unversioned:
 
2460
                                new_executable = bool(
 
2461
                                    stat.S_ISREG(current_path_info[3].st_mode)
 
2462
                                    and stat.S_IEXEC & current_path_info[3].st_mode)
 
2463
                                try:
 
2464
                                    relpath_unicode = utf8_decode(current_path_info[0])[0]
 
2465
                                except UnicodeDecodeError:
 
2466
                                    raise errors.BadFilenameEncoding(
 
2467
                                        current_path_info[0], osutils._fs_enc)
 
2468
                                yield (None,
 
2469
                                    (None, relpath_unicode),
 
2470
                                    True,
 
2471
                                    (False, False),
 
2472
                                    (None, None),
 
2473
                                    (None, utf8_decode(current_path_info[1])[0]),
 
2474
                                    (None, current_path_info[2]),
 
2475
                                    (None, new_executable))
 
2476
                            # dont descend into this unversioned path if it is
 
2477
                            # a dir
 
2478
                            if current_path_info[2] in ('directory'):
 
2479
                                del current_dir_info[1][path_index]
 
2480
                                path_index -= 1
 
2481
                        # dont descend the disk iterator into any tree 
 
2482
                        # paths.
 
2483
                        if current_path_info[2] == 'tree-reference':
 
2484
                            del current_dir_info[1][path_index]
 
2485
                            path_index -= 1
 
2486
                        path_index += 1
 
2487
                        if path_index < len(current_dir_info[1]):
 
2488
                            current_path_info = current_dir_info[1][path_index]
 
2489
                            if current_path_info[2] == 'directory':
 
2490
                                if self.target._directory_is_tree_reference(
 
2491
                                    current_path_info[0].decode('utf8')):
 
2492
                                    current_path_info = current_path_info[:2] + \
 
2493
                                        ('tree-reference',) + current_path_info[3:]
 
2494
                        else:
 
2495
                            current_path_info = None
 
2496
                        path_handled = False
 
2497
                    else:
 
2498
                        advance_path = True # reset the advance flagg.
 
2499
                if current_block is not None:
 
2500
                    block_index += 1
 
2501
                    if (block_index < len(state._dirblocks) and
 
2502
                        osutils.is_inside(current_root, state._dirblocks[block_index][0])):
 
2503
                        current_block = state._dirblocks[block_index]
 
2504
                    else:
 
2505
                        current_block = None
 
2506
                if current_dir_info is not None:
 
2507
                    try:
 
2508
                        current_dir_info = dir_iterator.next()
 
2509
                    except StopIteration:
 
2510
                        current_dir_info = None
 
2511
 
 
2512
 
 
2513
    @staticmethod
 
2514
    def is_compatible(source, target):
 
2515
        # the target must be a dirstate working tree
 
2516
        if not isinstance(target, WorkingTree4):
 
2517
            return False
 
2518
        # the source must be a revtreee or dirstate rev tree.
 
2519
        if not isinstance(source,
 
2520
            (revisiontree.RevisionTree, DirStateRevisionTree)):
 
2521
            return False
 
2522
        # the source revid must be in the target dirstate
 
2523
        if not (source._revision_id == NULL_REVISION or
 
2524
            source._revision_id in target.get_parent_ids()):
 
2525
            # TODO: what about ghosts? it may well need to 
 
2526
            # check for them explicitly.
 
2527
            return False
 
2528
        return True
 
2529
 
 
2530
InterTree.register_optimiser(InterDirStateTree)
 
2531
 
 
2532
 
 
2533
class Converter3to4(object):
 
2534
    """Perform an in-place upgrade of format 3 to format 4 trees."""
 
2535
 
 
2536
    def __init__(self):
 
2537
        self.target_format = WorkingTreeFormat4()
 
2538
 
 
2539
    def convert(self, tree):
 
2540
        # lock the control files not the tree, so that we dont get tree
 
2541
        # on-unlock behaviours, and so that noone else diddles with the 
 
2542
        # tree during upgrade.
 
2543
        tree._control_files.lock_write()
 
2544
        try:
 
2545
            tree.read_working_inventory()
 
2546
            self.create_dirstate_data(tree)
 
2547
            self.update_format(tree)
 
2548
            self.remove_xml_files(tree)
 
2549
        finally:
 
2550
            tree._control_files.unlock()
 
2551
 
 
2552
    def create_dirstate_data(self, tree):
 
2553
        """Create the dirstate based data for tree."""
 
2554
        local_path = tree.bzrdir.get_workingtree_transport(None
 
2555
            ).local_abspath('dirstate')
 
2556
        state = dirstate.DirState.from_tree(tree, local_path)
 
2557
        state.save()
 
2558
        state.unlock()
 
2559
 
 
2560
    def remove_xml_files(self, tree):
 
2561
        """Remove the oldformat 3 data."""
 
2562
        transport = tree.bzrdir.get_workingtree_transport(None)
 
2563
        for path in ['basis-inventory-cache', 'inventory', 'last-revision',
 
2564
            'pending-merges', 'stat-cache']:
 
2565
            try:
 
2566
                transport.delete(path)
 
2567
            except errors.NoSuchFile:
 
2568
                # some files are optional - just deal.
 
2569
                pass
 
2570
 
 
2571
    def update_format(self, tree):
 
2572
        """Change the format marker."""
 
2573
        tree._transport.put_bytes('format',
 
2574
            self.target_format.get_format_string(),
 
2575
            mode=tree.bzrdir._get_file_mode())