~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to bzrlib/workingtree_4.py

  • Committer: Alexander Belchenko
  • Date: 2008-03-11 08:49:42 UTC
  • mto: This revision was merged to the branch mainline in revision 3268.
  • Revision ID: bialix@ukr.net-20080311084942-w1w0w3v0m20p2pbc
use sys.version_info

Show diffs side-by-side

added added

removed removed

Lines of Context:
 
1
# Copyright (C) 2005, 2006, 2007 Canonical Ltd
 
2
#
 
3
# This program is free software; you can redistribute it and/or modify
 
4
# it under the terms of the GNU General Public License as published by
 
5
# the Free Software Foundation; either version 2 of the License, or
 
6
# (at your option) any later version.
 
7
#
 
8
# This program is distributed in the hope that it will be useful,
 
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
 
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 
11
# GNU General Public License for more details.
 
12
#
 
13
# You should have received a copy of the GNU General Public License
 
14
# along with this program; if not, write to the Free Software
 
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
 
16
 
 
17
"""WorkingTree4 format and implementation.
 
18
 
 
19
WorkingTree4 provides the dirstate based working tree logic.
 
20
 
 
21
To get a WorkingTree, call bzrdir.open_workingtree() or
 
22
WorkingTree.open(dir).
 
23
"""
 
24
 
 
25
from cStringIO import StringIO
 
26
import os
 
27
import sys
 
28
 
 
29
from bzrlib.lazy_import import lazy_import
 
30
lazy_import(globals(), """
 
31
from bisect import bisect_left
 
32
import collections
 
33
from copy import deepcopy
 
34
import errno
 
35
import itertools
 
36
import operator
 
37
import stat
 
38
from time import time
 
39
import warnings
 
40
 
 
41
import bzrlib
 
42
from bzrlib import (
 
43
    bzrdir,
 
44
    cache_utf8,
 
45
    conflicts as _mod_conflicts,
 
46
    debug,
 
47
    delta,
 
48
    dirstate,
 
49
    errors,
 
50
    generate_ids,
 
51
    globbing,
 
52
    ignores,
 
53
    merge,
 
54
    osutils,
 
55
    revision as _mod_revision,
 
56
    revisiontree,
 
57
    textui,
 
58
    trace,
 
59
    transform,
 
60
    urlutils,
 
61
    xml5,
 
62
    xml6,
 
63
    )
 
64
import bzrlib.branch
 
65
from bzrlib.transport import get_transport
 
66
import bzrlib.ui
 
67
""")
 
68
 
 
69
from bzrlib import symbol_versioning
 
70
from bzrlib.decorators import needs_read_lock, needs_write_lock
 
71
from bzrlib.inventory import InventoryEntry, Inventory, ROOT_ID, entry_factory
 
72
from bzrlib.lockable_files import LockableFiles, TransportLock
 
73
from bzrlib.lockdir import LockDir
 
74
import bzrlib.mutabletree
 
75
from bzrlib.mutabletree import needs_tree_write_lock
 
76
from bzrlib.osutils import (
 
77
    file_kind,
 
78
    isdir,
 
79
    normpath,
 
80
    pathjoin,
 
81
    rand_chars,
 
82
    realpath,
 
83
    safe_unicode,
 
84
    splitpath,
 
85
    )
 
86
from bzrlib.trace import mutter, note
 
87
from bzrlib.transport.local import LocalTransport
 
88
from bzrlib.tree import InterTree
 
89
from bzrlib.progress import DummyProgress, ProgressPhase
 
90
from bzrlib.revision import NULL_REVISION, CURRENT_REVISION
 
91
from bzrlib.rio import RioReader, rio_file, Stanza
 
92
from bzrlib.symbol_versioning import (deprecated_passed,
 
93
        deprecated_method,
 
94
        deprecated_function,
 
95
        DEPRECATED_PARAMETER,
 
96
        )
 
97
from bzrlib.tree import Tree
 
98
from bzrlib.workingtree import WorkingTree, WorkingTree3, WorkingTreeFormat3
 
99
 
 
100
 
 
101
# This is the Windows equivalent of ENOTDIR
 
102
# It is defined in pywin32.winerror, but we don't want a strong dependency for
 
103
# just an error code.
 
104
ERROR_PATH_NOT_FOUND = 3
 
105
ERROR_DIRECTORY = 267
 
106
 
 
107
 
 
108
class WorkingTree4(WorkingTree3):
 
109
    """This is the Format 4 working tree.
 
110
 
 
111
    This differs from WorkingTree3 by:
 
112
     - Having a consolidated internal dirstate, stored in a
 
113
       randomly-accessible sorted file on disk.
 
114
     - Not having a regular inventory attribute.  One can be synthesized 
 
115
       on demand but this is expensive and should be avoided.
 
116
 
 
117
    This is new in bzr 0.15.
 
118
    """
 
119
 
 
120
    def __init__(self, basedir,
 
121
                 branch,
 
122
                 _control_files=None,
 
123
                 _format=None,
 
124
                 _bzrdir=None):
 
125
        """Construct a WorkingTree for basedir.
 
126
 
 
127
        If the branch is not supplied, it is opened automatically.
 
128
        If the branch is supplied, it must be the branch for this basedir.
 
129
        (branch.base is not cross checked, because for remote branches that
 
130
        would be meaningless).
 
131
        """
 
132
        self._format = _format
 
133
        self.bzrdir = _bzrdir
 
134
        assert isinstance(basedir, basestring), \
 
135
            "base directory %r is not a string" % basedir
 
136
        basedir = safe_unicode(basedir)
 
137
        mutter("opening working tree %r", basedir)
 
138
        self._branch = branch
 
139
        assert isinstance(self.branch, bzrlib.branch.Branch), \
 
140
            "branch %r is not a Branch" % self.branch
 
141
        self.basedir = realpath(basedir)
 
142
        # if branch is at our basedir and is a format 6 or less
 
143
        # assume all other formats have their own control files.
 
144
        assert isinstance(_control_files, LockableFiles), \
 
145
            "_control_files must be a LockableFiles, not %r" % _control_files
 
146
        self._control_files = _control_files
 
147
        self._dirty = None
 
148
        #-------------
 
149
        # during a read or write lock these objects are set, and are
 
150
        # None the rest of the time.
 
151
        self._dirstate = None
 
152
        self._inventory = None
 
153
        #-------------
 
154
        self._setup_directory_is_tree_reference()
 
155
        self._detect_case_handling()
 
156
 
 
157
    @needs_tree_write_lock
 
158
    def _add(self, files, ids, kinds):
 
159
        """See MutableTree._add."""
 
160
        state = self.current_dirstate()
 
161
        for f, file_id, kind in zip(files, ids, kinds):
 
162
            f = f.strip('/')
 
163
            if self.path2id(f):
 
164
                # special case tree root handling.
 
165
                if f == '' and self.path2id(f) == ROOT_ID:
 
166
                    state.set_path_id('', generate_ids.gen_file_id(f))
 
167
                continue
 
168
            if file_id is None:
 
169
                file_id = generate_ids.gen_file_id(f)
 
170
            # deliberately add the file with no cached stat or sha1
 
171
            # - on the first access it will be gathered, and we can
 
172
            # always change this once tests are all passing.
 
173
            state.add(f, file_id, kind, None, '')
 
174
        self._make_dirty(reset_inventory=True)
 
175
 
 
176
    def _make_dirty(self, reset_inventory):
 
177
        """Make the tree state dirty.
 
178
 
 
179
        :param reset_inventory: True if the cached inventory should be removed
 
180
            (presuming there is one).
 
181
        """
 
182
        self._dirty = True
 
183
        if reset_inventory and self._inventory is not None:
 
184
            self._inventory = None
 
185
 
 
186
    @needs_tree_write_lock
 
187
    def add_reference(self, sub_tree):
 
188
        # use standard implementation, which calls back to self._add
 
189
        # 
 
190
        # So we don't store the reference_revision in the working dirstate,
 
191
        # it's just recorded at the moment of commit. 
 
192
        self._add_reference(sub_tree)
 
193
 
 
194
    def break_lock(self):
 
195
        """Break a lock if one is present from another instance.
 
196
 
 
197
        Uses the ui factory to ask for confirmation if the lock may be from
 
198
        an active process.
 
199
 
 
200
        This will probe the repository for its lock as well.
 
201
        """
 
202
        # if the dirstate is locked by an active process, reject the break lock
 
203
        # call.
 
204
        try:
 
205
            if self._dirstate is None:
 
206
                clear = True
 
207
            else:
 
208
                clear = False
 
209
            state = self._current_dirstate()
 
210
            if state._lock_token is not None:
 
211
                # we already have it locked. sheese, cant break our own lock.
 
212
                raise errors.LockActive(self.basedir)
 
213
            else:
 
214
                try:
 
215
                    # try for a write lock - need permission to get one anyhow
 
216
                    # to break locks.
 
217
                    state.lock_write()
 
218
                except errors.LockContention:
 
219
                    # oslocks fail when a process is still live: fail.
 
220
                    # TODO: get the locked lockdir info and give to the user to
 
221
                    # assist in debugging.
 
222
                    raise errors.LockActive(self.basedir)
 
223
                else:
 
224
                    state.unlock()
 
225
        finally:
 
226
            if clear:
 
227
                self._dirstate = None
 
228
        self._control_files.break_lock()
 
229
        self.branch.break_lock()
 
230
 
 
231
    def _comparison_data(self, entry, path):
 
232
        kind, executable, stat_value = \
 
233
            WorkingTree3._comparison_data(self, entry, path)
 
234
        # it looks like a plain directory, but it's really a reference -- see
 
235
        # also kind()
 
236
        if (self._repo_supports_tree_reference and
 
237
            kind == 'directory' and
 
238
            self._directory_is_tree_reference(path)):
 
239
            kind = 'tree-reference'
 
240
        return kind, executable, stat_value
 
241
 
 
242
    @needs_write_lock
 
243
    def commit(self, message=None, revprops=None, *args, **kwargs):
 
244
        # mark the tree as dirty post commit - commit
 
245
        # can change the current versioned list by doing deletes.
 
246
        result = WorkingTree3.commit(self, message, revprops, *args, **kwargs)
 
247
        self._make_dirty(reset_inventory=True)
 
248
        return result
 
249
 
 
250
    def current_dirstate(self):
 
251
        """Return the current dirstate object.
 
252
 
 
253
        This is not part of the tree interface and only exposed for ease of
 
254
        testing.
 
255
 
 
256
        :raises errors.NotWriteLocked: when not in a lock.
 
257
        """
 
258
        self._must_be_locked()
 
259
        return self._current_dirstate()
 
260
 
 
261
    def _current_dirstate(self):
 
262
        """Internal function that does not check lock status.
 
263
 
 
264
        This is needed for break_lock which also needs the dirstate.
 
265
        """
 
266
        if self._dirstate is not None:
 
267
            return self._dirstate
 
268
        local_path = self.bzrdir.get_workingtree_transport(None
 
269
            ).local_abspath('dirstate')
 
270
        self._dirstate = dirstate.DirState.on_file(local_path)
 
271
        return self._dirstate
 
272
 
 
273
    def filter_unversioned_files(self, paths):
 
274
        """Filter out paths that are versioned.
 
275
 
 
276
        :return: set of paths.
 
277
        """
 
278
        # TODO: make a generic multi-bisect routine roughly that should list
 
279
        # the paths, then process one half at a time recursively, and feed the
 
280
        # results of each bisect in further still
 
281
        paths = sorted(paths)
 
282
        result = set()
 
283
        state = self.current_dirstate()
 
284
        # TODO we want a paths_to_dirblocks helper I think
 
285
        for path in paths:
 
286
            dirname, basename = os.path.split(path.encode('utf8'))
 
287
            _, _, _, path_is_versioned = state._get_block_entry_index(
 
288
                dirname, basename, 0)
 
289
            if not path_is_versioned:
 
290
                result.add(path)
 
291
        return result
 
292
 
 
293
    def flush(self):
 
294
        """Write all cached data to disk."""
 
295
        if self._control_files._lock_mode != 'w':
 
296
            raise errors.NotWriteLocked(self)
 
297
        self.current_dirstate().save()
 
298
        self._inventory = None
 
299
        self._dirty = False
 
300
 
 
301
    @needs_tree_write_lock
 
302
    def _gather_kinds(self, files, kinds):
 
303
        """See MutableTree._gather_kinds."""
 
304
        for pos, f in enumerate(files):
 
305
            if kinds[pos] is None:
 
306
                kinds[pos] = self._kind(f)
 
307
 
 
308
    def _generate_inventory(self):
 
309
        """Create and set self.inventory from the dirstate object.
 
310
        
 
311
        This is relatively expensive: we have to walk the entire dirstate.
 
312
        Ideally we would not, and can deprecate this function.
 
313
        """
 
314
        #: uncomment to trap on inventory requests.
 
315
        # import pdb;pdb.set_trace()
 
316
        state = self.current_dirstate()
 
317
        state._read_dirblocks_if_needed()
 
318
        root_key, current_entry = self._get_entry(path='')
 
319
        current_id = root_key[2]
 
320
        assert current_entry[0][0] == 'd' # directory
 
321
        inv = Inventory(root_id=current_id)
 
322
        # Turn some things into local variables
 
323
        minikind_to_kind = dirstate.DirState._minikind_to_kind
 
324
        factory = entry_factory
 
325
        utf8_decode = cache_utf8._utf8_decode
 
326
        inv_byid = inv._byid
 
327
        # we could do this straight out of the dirstate; it might be fast
 
328
        # and should be profiled - RBC 20070216
 
329
        parent_ies = {'' : inv.root}
 
330
        for block in state._dirblocks[1:]: # skip the root
 
331
            dirname = block[0]
 
332
            try:
 
333
                parent_ie = parent_ies[dirname]
 
334
            except KeyError:
 
335
                # all the paths in this block are not versioned in this tree
 
336
                continue
 
337
            for key, entry in block[1]:
 
338
                minikind, link_or_sha1, size, executable, stat = entry[0]
 
339
                if minikind in ('a', 'r'): # absent, relocated
 
340
                    # a parent tree only entry
 
341
                    continue
 
342
                name = key[1]
 
343
                name_unicode = utf8_decode(name)[0]
 
344
                file_id = key[2]
 
345
                kind = minikind_to_kind[minikind]
 
346
                inv_entry = factory[kind](file_id, name_unicode,
 
347
                                          parent_ie.file_id)
 
348
                if kind == 'file':
 
349
                    # This is only needed on win32, where this is the only way
 
350
                    # we know the executable bit.
 
351
                    inv_entry.executable = executable
 
352
                    # not strictly needed: working tree
 
353
                    #inv_entry.text_size = size
 
354
                    #inv_entry.text_sha1 = sha1
 
355
                elif kind == 'directory':
 
356
                    # add this entry to the parent map.
 
357
                    parent_ies[(dirname + '/' + name).strip('/')] = inv_entry
 
358
                elif kind == 'tree-reference':
 
359
                    assert self._repo_supports_tree_reference, \
 
360
                        "repository of %r " \
 
361
                        "doesn't support tree references " \
 
362
                        "required by entry %r" \
 
363
                        % (self, name)
 
364
                    inv_entry.reference_revision = link_or_sha1 or None
 
365
                elif kind != 'symlink':
 
366
                    raise AssertionError("unknown kind %r" % kind)
 
367
                # These checks cost us around 40ms on a 55k entry tree
 
368
                assert file_id not in inv_byid, ('file_id %s already in'
 
369
                    ' inventory as %s' % (file_id, inv_byid[file_id]))
 
370
                assert name_unicode not in parent_ie.children
 
371
                inv_byid[file_id] = inv_entry
 
372
                parent_ie.children[name_unicode] = inv_entry
 
373
        self._inventory = inv
 
374
 
 
375
    def _get_entry(self, file_id=None, path=None):
 
376
        """Get the dirstate row for file_id or path.
 
377
 
 
378
        If either file_id or path is supplied, it is used as the key to lookup.
 
379
        If both are supplied, the fastest lookup is used, and an error is
 
380
        raised if they do not both point at the same row.
 
381
        
 
382
        :param file_id: An optional unicode file_id to be looked up.
 
383
        :param path: An optional unicode path to be looked up.
 
384
        :return: The dirstate row tuple for path/file_id, or (None, None)
 
385
        """
 
386
        if file_id is None and path is None:
 
387
            raise errors.BzrError('must supply file_id or path')
 
388
        state = self.current_dirstate()
 
389
        if path is not None:
 
390
            path = path.encode('utf8')
 
391
        return state._get_entry(0, fileid_utf8=file_id, path_utf8=path)
 
392
 
 
393
    def get_file_sha1(self, file_id, path=None, stat_value=None):
 
394
        # check file id is valid unconditionally.
 
395
        entry = self._get_entry(file_id=file_id, path=path)
 
396
        if entry[0] is None:
 
397
            raise errors.NoSuchId(self, file_id)
 
398
        if path is None:
 
399
            path = pathjoin(entry[0][0], entry[0][1]).decode('utf8')
 
400
 
 
401
        file_abspath = self.abspath(path)
 
402
        state = self.current_dirstate()
 
403
        if stat_value is None:
 
404
            try:
 
405
                stat_value = os.lstat(file_abspath)
 
406
            except OSError, e:
 
407
                if e.errno == errno.ENOENT:
 
408
                    return None
 
409
                else:
 
410
                    raise
 
411
        link_or_sha1 = state.update_entry(entry, file_abspath,
 
412
                                          stat_value=stat_value)
 
413
        if entry[1][0][0] == 'f':
 
414
            return link_or_sha1
 
415
        return None
 
416
 
 
417
    def _get_inventory(self):
 
418
        """Get the inventory for the tree. This is only valid within a lock."""
 
419
        if 'evil' in debug.debug_flags:
 
420
            trace.mutter_callsite(2,
 
421
                "accessing .inventory forces a size of tree translation.")
 
422
        if self._inventory is not None:
 
423
            return self._inventory
 
424
        self._must_be_locked()
 
425
        self._generate_inventory()
 
426
        return self._inventory
 
427
 
 
428
    inventory = property(_get_inventory,
 
429
                         doc="Inventory of this Tree")
 
430
 
 
431
    @needs_read_lock
 
432
    def get_parent_ids(self):
 
433
        """See Tree.get_parent_ids.
 
434
        
 
435
        This implementation requests the ids list from the dirstate file.
 
436
        """
 
437
        return self.current_dirstate().get_parent_ids()
 
438
 
 
439
    def get_reference_revision(self, file_id, path=None):
 
440
        # referenced tree's revision is whatever's currently there
 
441
        return self.get_nested_tree(file_id, path).last_revision()
 
442
 
 
443
    def get_nested_tree(self, file_id, path=None):
 
444
        if path is None:
 
445
            path = self.id2path(file_id)
 
446
        # else: check file_id is at path?
 
447
        return WorkingTree.open(self.abspath(path))
 
448
 
 
449
    @needs_read_lock
 
450
    def get_root_id(self):
 
451
        """Return the id of this trees root"""
 
452
        return self._get_entry(path='')[0][2]
 
453
 
 
454
    def has_id(self, file_id):
 
455
        state = self.current_dirstate()
 
456
        row, parents = self._get_entry(file_id=file_id)
 
457
        if row is None:
 
458
            return False
 
459
        return osutils.lexists(pathjoin(
 
460
                    self.basedir, row[0].decode('utf8'), row[1].decode('utf8')))
 
461
 
 
462
    @needs_read_lock
 
463
    def id2path(self, file_id):
 
464
        "Convert a file-id to a path."
 
465
        state = self.current_dirstate()
 
466
        entry = self._get_entry(file_id=file_id)
 
467
        if entry == (None, None):
 
468
            raise errors.NoSuchId(tree=self, file_id=file_id)
 
469
        path_utf8 = osutils.pathjoin(entry[0][0], entry[0][1])
 
470
        return path_utf8.decode('utf8')
 
471
 
 
472
    def _is_executable_from_path_and_stat_from_basis(self, path, stat_result):
 
473
        entry = self._get_entry(path=path)
 
474
        if entry == (None, None):
 
475
            return False # Missing entries are not executable
 
476
        return entry[1][0][3] # Executable?
 
477
 
 
478
    if not osutils.supports_executable():
 
479
        def is_executable(self, file_id, path=None):
 
480
            """Test if a file is executable or not.
 
481
 
 
482
            Note: The caller is expected to take a read-lock before calling this.
 
483
            """
 
484
            entry = self._get_entry(file_id=file_id, path=path)
 
485
            if entry == (None, None):
 
486
                return False
 
487
            return entry[1][0][3]
 
488
 
 
489
        _is_executable_from_path_and_stat = \
 
490
            _is_executable_from_path_and_stat_from_basis
 
491
    else:
 
492
        def is_executable(self, file_id, path=None):
 
493
            """Test if a file is executable or not.
 
494
 
 
495
            Note: The caller is expected to take a read-lock before calling this.
 
496
            """
 
497
            self._must_be_locked()
 
498
            if not path:
 
499
                path = self.id2path(file_id)
 
500
            mode = os.lstat(self.abspath(path)).st_mode
 
501
            return bool(stat.S_ISREG(mode) and stat.S_IEXEC & mode)
 
502
 
 
503
    def all_file_ids(self):
 
504
        """See Tree.iter_all_file_ids"""
 
505
        self._must_be_locked()
 
506
        result = set()
 
507
        for key, tree_details in self.current_dirstate()._iter_entries():
 
508
            if tree_details[0][0] in ('a', 'r'): # relocated
 
509
                continue
 
510
            result.add(key[2])
 
511
        return result
 
512
 
 
513
    @needs_read_lock
 
514
    def __iter__(self):
 
515
        """Iterate through file_ids for this tree.
 
516
 
 
517
        file_ids are in a WorkingTree if they are in the working inventory
 
518
        and the working file exists.
 
519
        """
 
520
        result = []
 
521
        for key, tree_details in self.current_dirstate()._iter_entries():
 
522
            if tree_details[0][0] in ('a', 'r'): # absent, relocated
 
523
                # not relevant to the working tree
 
524
                continue
 
525
            path = pathjoin(self.basedir, key[0].decode('utf8'), key[1].decode('utf8'))
 
526
            if osutils.lexists(path):
 
527
                result.append(key[2])
 
528
        return iter(result)
 
529
 
 
530
    def iter_references(self):
 
531
        for key, tree_details in self.current_dirstate()._iter_entries():
 
532
            if tree_details[0][0] in ('a', 'r'): # absent, relocated
 
533
                # not relevant to the working tree
 
534
                continue
 
535
            if not key[1]:
 
536
                # the root is not a reference.
 
537
                continue
 
538
            path = pathjoin(self.basedir, key[0].decode('utf8'), key[1].decode('utf8'))
 
539
            try:
 
540
                if self._kind(path) == 'tree-reference':
 
541
                    yield path, key[2]
 
542
            except errors.NoSuchFile:
 
543
                # path is missing on disk.
 
544
                continue
 
545
 
 
546
    def kind(self, file_id):
 
547
        """Return the kind of a file.
 
548
 
 
549
        This is always the actual kind that's on disk, regardless of what it
 
550
        was added as.
 
551
 
 
552
        Note: The caller is expected to take a read-lock before calling this.
 
553
        """
 
554
        relpath = self.id2path(file_id)
 
555
        assert relpath != None, \
 
556
            "path for id {%s} is None!" % file_id
 
557
        return self._kind(relpath)
 
558
 
 
559
    def _kind(self, relpath):
 
560
        abspath = self.abspath(relpath)
 
561
        kind = file_kind(abspath)
 
562
        if (self._repo_supports_tree_reference and
 
563
            kind == 'directory' and
 
564
            self._directory_is_tree_reference(relpath)):
 
565
            kind = 'tree-reference'
 
566
        return kind
 
567
 
 
568
    @needs_read_lock
 
569
    def _last_revision(self):
 
570
        """See Mutable.last_revision."""
 
571
        parent_ids = self.current_dirstate().get_parent_ids()
 
572
        if parent_ids:
 
573
            return parent_ids[0]
 
574
        else:
 
575
            return _mod_revision.NULL_REVISION
 
576
 
 
577
    def lock_read(self):
 
578
        """See Branch.lock_read, and WorkingTree.unlock."""
 
579
        self.branch.lock_read()
 
580
        try:
 
581
            self._control_files.lock_read()
 
582
            try:
 
583
                state = self.current_dirstate()
 
584
                if not state._lock_token:
 
585
                    state.lock_read()
 
586
                # set our support for tree references from the repository in
 
587
                # use.
 
588
                self._repo_supports_tree_reference = getattr(
 
589
                    self.branch.repository._format, "supports_tree_reference",
 
590
                    False)
 
591
            except:
 
592
                self._control_files.unlock()
 
593
                raise
 
594
        except:
 
595
            self.branch.unlock()
 
596
            raise
 
597
 
 
598
    def _lock_self_write(self):
 
599
        """This should be called after the branch is locked."""
 
600
        try:
 
601
            self._control_files.lock_write()
 
602
            try:
 
603
                state = self.current_dirstate()
 
604
                if not state._lock_token:
 
605
                    state.lock_write()
 
606
                # set our support for tree references from the repository in
 
607
                # use.
 
608
                self._repo_supports_tree_reference = getattr(
 
609
                    self.branch.repository._format, "supports_tree_reference",
 
610
                    False)
 
611
            except:
 
612
                self._control_files.unlock()
 
613
                raise
 
614
        except:
 
615
            self.branch.unlock()
 
616
            raise
 
617
 
 
618
    def lock_tree_write(self):
 
619
        """See MutableTree.lock_tree_write, and WorkingTree.unlock."""
 
620
        self.branch.lock_read()
 
621
        self._lock_self_write()
 
622
 
 
623
    def lock_write(self):
 
624
        """See MutableTree.lock_write, and WorkingTree.unlock."""
 
625
        self.branch.lock_write()
 
626
        self._lock_self_write()
 
627
 
 
628
    @needs_tree_write_lock
 
629
    def move(self, from_paths, to_dir, after=False):
 
630
        """See WorkingTree.move()."""
 
631
        result = []
 
632
        if not from_paths:
 
633
            return result
 
634
 
 
635
        state = self.current_dirstate()
 
636
 
 
637
        assert not isinstance(from_paths, basestring)
 
638
        to_dir_utf8 = to_dir.encode('utf8')
 
639
        to_entry_dirname, to_basename = os.path.split(to_dir_utf8)
 
640
        id_index = state._get_id_index()
 
641
        # check destination directory
 
642
        # get the details for it
 
643
        to_entry_block_index, to_entry_entry_index, dir_present, entry_present = \
 
644
            state._get_block_entry_index(to_entry_dirname, to_basename, 0)
 
645
        if not entry_present:
 
646
            raise errors.BzrMoveFailedError('', to_dir,
 
647
                errors.NotVersionedError(to_dir))
 
648
        to_entry = state._dirblocks[to_entry_block_index][1][to_entry_entry_index]
 
649
        # get a handle on the block itself.
 
650
        to_block_index = state._ensure_block(
 
651
            to_entry_block_index, to_entry_entry_index, to_dir_utf8)
 
652
        to_block = state._dirblocks[to_block_index]
 
653
        to_abs = self.abspath(to_dir)
 
654
        if not isdir(to_abs):
 
655
            raise errors.BzrMoveFailedError('',to_dir,
 
656
                errors.NotADirectory(to_abs))
 
657
 
 
658
        if to_entry[1][0][0] != 'd':
 
659
            raise errors.BzrMoveFailedError('',to_dir,
 
660
                errors.NotADirectory(to_abs))
 
661
 
 
662
        if self._inventory is not None:
 
663
            update_inventory = True
 
664
            inv = self.inventory
 
665
            to_dir_ie = inv[to_dir_id]
 
666
            to_dir_id = to_entry[0][2]
 
667
        else:
 
668
            update_inventory = False
 
669
 
 
670
        rollbacks = []
 
671
        def move_one(old_entry, from_path_utf8, minikind, executable,
 
672
                     fingerprint, packed_stat, size,
 
673
                     to_block, to_key, to_path_utf8):
 
674
            state._make_absent(old_entry)
 
675
            from_key = old_entry[0]
 
676
            rollbacks.append(
 
677
                lambda:state.update_minimal(from_key,
 
678
                    minikind,
 
679
                    executable=executable,
 
680
                    fingerprint=fingerprint,
 
681
                    packed_stat=packed_stat,
 
682
                    size=size,
 
683
                    path_utf8=from_path_utf8))
 
684
            state.update_minimal(to_key,
 
685
                    minikind,
 
686
                    executable=executable,
 
687
                    fingerprint=fingerprint,
 
688
                    packed_stat=packed_stat,
 
689
                    size=size,
 
690
                    path_utf8=to_path_utf8)
 
691
            added_entry_index, _ = state._find_entry_index(to_key, to_block[1])
 
692
            new_entry = to_block[1][added_entry_index]
 
693
            rollbacks.append(lambda:state._make_absent(new_entry))
 
694
 
 
695
        for from_rel in from_paths:
 
696
            # from_rel is 'pathinroot/foo/bar'
 
697
            from_rel_utf8 = from_rel.encode('utf8')
 
698
            from_dirname, from_tail = osutils.split(from_rel)
 
699
            from_dirname, from_tail_utf8 = osutils.split(from_rel_utf8)
 
700
            from_entry = self._get_entry(path=from_rel)
 
701
            if from_entry == (None, None):
 
702
                raise errors.BzrMoveFailedError(from_rel,to_dir,
 
703
                    errors.NotVersionedError(path=str(from_rel)))
 
704
 
 
705
            from_id = from_entry[0][2]
 
706
            to_rel = pathjoin(to_dir, from_tail)
 
707
            to_rel_utf8 = pathjoin(to_dir_utf8, from_tail_utf8)
 
708
            item_to_entry = self._get_entry(path=to_rel)
 
709
            if item_to_entry != (None, None):
 
710
                raise errors.BzrMoveFailedError(from_rel, to_rel,
 
711
                    "Target is already versioned.")
 
712
 
 
713
            if from_rel == to_rel:
 
714
                raise errors.BzrMoveFailedError(from_rel, to_rel,
 
715
                    "Source and target are identical.")
 
716
 
 
717
            from_missing = not self.has_filename(from_rel)
 
718
            to_missing = not self.has_filename(to_rel)
 
719
            if after:
 
720
                move_file = False
 
721
            else:
 
722
                move_file = True
 
723
            if to_missing:
 
724
                if not move_file:
 
725
                    raise errors.BzrMoveFailedError(from_rel, to_rel,
 
726
                        errors.NoSuchFile(path=to_rel,
 
727
                        extra="New file has not been created yet"))
 
728
                elif from_missing:
 
729
                    # neither path exists
 
730
                    raise errors.BzrRenameFailedError(from_rel, to_rel,
 
731
                        errors.PathsDoNotExist(paths=(from_rel, to_rel)))
 
732
            else:
 
733
                if from_missing: # implicitly just update our path mapping
 
734
                    move_file = False
 
735
                elif not after:
 
736
                    raise errors.RenameFailedFilesExist(from_rel, to_rel)
 
737
 
 
738
            rollbacks = []
 
739
            def rollback_rename():
 
740
                """A single rename has failed, roll it back."""
 
741
                # roll back everything, even if we encounter trouble doing one
 
742
                # of them.
 
743
                #
 
744
                # TODO: at least log the other exceptions rather than just
 
745
                # losing them mbp 20070307
 
746
                exc_info = None
 
747
                for rollback in reversed(rollbacks):
 
748
                    try:
 
749
                        rollback()
 
750
                    except Exception, e:
 
751
                        exc_info = sys.exc_info()
 
752
                if exc_info:
 
753
                    raise exc_info[0], exc_info[1], exc_info[2]
 
754
 
 
755
            # perform the disk move first - its the most likely failure point.
 
756
            if move_file:
 
757
                from_rel_abs = self.abspath(from_rel)
 
758
                to_rel_abs = self.abspath(to_rel)
 
759
                try:
 
760
                    osutils.rename(from_rel_abs, to_rel_abs)
 
761
                except OSError, e:
 
762
                    raise errors.BzrMoveFailedError(from_rel, to_rel, e[1])
 
763
                rollbacks.append(lambda: osutils.rename(to_rel_abs, from_rel_abs))
 
764
            try:
 
765
                # perform the rename in the inventory next if needed: its easy
 
766
                # to rollback
 
767
                if update_inventory:
 
768
                    # rename the entry
 
769
                    from_entry = inv[from_id]
 
770
                    current_parent = from_entry.parent_id
 
771
                    inv.rename(from_id, to_dir_id, from_tail)
 
772
                    rollbacks.append(
 
773
                        lambda: inv.rename(from_id, current_parent, from_tail))
 
774
                # finally do the rename in the dirstate, which is a little
 
775
                # tricky to rollback, but least likely to need it.
 
776
                old_block_index, old_entry_index, dir_present, file_present = \
 
777
                    state._get_block_entry_index(from_dirname, from_tail_utf8, 0)
 
778
                old_block = state._dirblocks[old_block_index][1]
 
779
                old_entry = old_block[old_entry_index]
 
780
                from_key, old_entry_details = old_entry
 
781
                cur_details = old_entry_details[0]
 
782
                # remove the old row
 
783
                to_key = ((to_block[0],) + from_key[1:3])
 
784
                minikind = cur_details[0]
 
785
                move_one(old_entry, from_path_utf8=from_rel_utf8,
 
786
                         minikind=minikind,
 
787
                         executable=cur_details[3],
 
788
                         fingerprint=cur_details[1],
 
789
                         packed_stat=cur_details[4],
 
790
                         size=cur_details[2],
 
791
                         to_block=to_block,
 
792
                         to_key=to_key,
 
793
                         to_path_utf8=to_rel_utf8)
 
794
 
 
795
                if minikind == 'd':
 
796
                    def update_dirblock(from_dir, to_key, to_dir_utf8):
 
797
                        """Recursively update all entries in this dirblock."""
 
798
                        assert from_dir != '', "renaming root not supported"
 
799
                        from_key = (from_dir, '')
 
800
                        from_block_idx, present = \
 
801
                            state._find_block_index_from_key(from_key)
 
802
                        if not present:
 
803
                            # This is the old record, if it isn't present, then
 
804
                            # there is theoretically nothing to update.
 
805
                            # (Unless it isn't present because of lazy loading,
 
806
                            # but we don't do that yet)
 
807
                            return
 
808
                        from_block = state._dirblocks[from_block_idx]
 
809
                        to_block_index, to_entry_index, _, _ = \
 
810
                            state._get_block_entry_index(to_key[0], to_key[1], 0)
 
811
                        to_block_index = state._ensure_block(
 
812
                            to_block_index, to_entry_index, to_dir_utf8)
 
813
                        to_block = state._dirblocks[to_block_index]
 
814
 
 
815
                        # Grab a copy since move_one may update the list.
 
816
                        for entry in from_block[1][:]:
 
817
                            assert entry[0][0] == from_dir
 
818
                            cur_details = entry[1][0]
 
819
                            to_key = (to_dir_utf8, entry[0][1], entry[0][2])
 
820
                            from_path_utf8 = osutils.pathjoin(entry[0][0], entry[0][1])
 
821
                            to_path_utf8 = osutils.pathjoin(to_dir_utf8, entry[0][1])
 
822
                            minikind = cur_details[0]
 
823
                            if minikind in 'ar':
 
824
                                # Deleted children of a renamed directory
 
825
                                # Do not need to be updated.
 
826
                                # Children that have been renamed out of this
 
827
                                # directory should also not be updated
 
828
                                continue
 
829
                            move_one(entry, from_path_utf8=from_path_utf8,
 
830
                                     minikind=minikind,
 
831
                                     executable=cur_details[3],
 
832
                                     fingerprint=cur_details[1],
 
833
                                     packed_stat=cur_details[4],
 
834
                                     size=cur_details[2],
 
835
                                     to_block=to_block,
 
836
                                     to_key=to_key,
 
837
                                     to_path_utf8=to_path_utf8)
 
838
                            if minikind == 'd':
 
839
                                # We need to move all the children of this
 
840
                                # entry
 
841
                                update_dirblock(from_path_utf8, to_key,
 
842
                                                to_path_utf8)
 
843
                    update_dirblock(from_rel_utf8, to_key, to_rel_utf8)
 
844
            except:
 
845
                rollback_rename()
 
846
                raise
 
847
            result.append((from_rel, to_rel))
 
848
            state._dirblock_state = dirstate.DirState.IN_MEMORY_MODIFIED
 
849
            self._make_dirty(reset_inventory=False)
 
850
 
 
851
        return result
 
852
 
 
853
    def _must_be_locked(self):
 
854
        if not self._control_files._lock_count:
 
855
            raise errors.ObjectNotLocked(self)
 
856
 
 
857
    def _new_tree(self):
 
858
        """Initialize the state in this tree to be a new tree."""
 
859
        self._dirty = True
 
860
 
 
861
    @needs_read_lock
 
862
    def path2id(self, path):
 
863
        """Return the id for path in this tree."""
 
864
        path = path.strip('/')
 
865
        entry = self._get_entry(path=path)
 
866
        if entry == (None, None):
 
867
            return None
 
868
        return entry[0][2]
 
869
 
 
870
    def paths2ids(self, paths, trees=[], require_versioned=True):
 
871
        """See Tree.paths2ids().
 
872
 
 
873
        This specialisation fast-paths the case where all the trees are in the
 
874
        dirstate.
 
875
        """
 
876
        if paths is None:
 
877
            return None
 
878
        parents = self.get_parent_ids()
 
879
        for tree in trees:
 
880
            if not (isinstance(tree, DirStateRevisionTree) and tree._revision_id in
 
881
                parents):
 
882
                return super(WorkingTree4, self).paths2ids(paths, trees, require_versioned)
 
883
        search_indexes = [0] + [1 + parents.index(tree._revision_id) for tree in trees]
 
884
        # -- make all paths utf8 --
 
885
        paths_utf8 = set()
 
886
        for path in paths:
 
887
            paths_utf8.add(path.encode('utf8'))
 
888
        paths = paths_utf8
 
889
        # -- paths is now a utf8 path set --
 
890
        # -- get the state object and prepare it.
 
891
        state = self.current_dirstate()
 
892
        if False and (state._dirblock_state == dirstate.DirState.NOT_IN_MEMORY
 
893
            and '' not in paths):
 
894
            paths2ids = self._paths2ids_using_bisect
 
895
        else:
 
896
            paths2ids = self._paths2ids_in_memory
 
897
        return paths2ids(paths, search_indexes,
 
898
                         require_versioned=require_versioned)
 
899
 
 
900
    def _paths2ids_in_memory(self, paths, search_indexes,
 
901
                             require_versioned=True):
 
902
        state = self.current_dirstate()
 
903
        state._read_dirblocks_if_needed()
 
904
        def _entries_for_path(path):
 
905
            """Return a list with all the entries that match path for all ids.
 
906
            """
 
907
            dirname, basename = os.path.split(path)
 
908
            key = (dirname, basename, '')
 
909
            block_index, present = state._find_block_index_from_key(key)
 
910
            if not present:
 
911
                # the block which should contain path is absent.
 
912
                return []
 
913
            result = []
 
914
            block = state._dirblocks[block_index][1]
 
915
            entry_index, _ = state._find_entry_index(key, block)
 
916
            # we may need to look at multiple entries at this path: walk while the paths match.
 
917
            while (entry_index < len(block) and
 
918
                block[entry_index][0][0:2] == key[0:2]):
 
919
                result.append(block[entry_index])
 
920
                entry_index += 1
 
921
            return result
 
922
        if require_versioned:
 
923
            # -- check all supplied paths are versioned in a search tree. --
 
924
            all_versioned = True
 
925
            for path in paths:
 
926
                path_entries = _entries_for_path(path)
 
927
                if not path_entries:
 
928
                    # this specified path is not present at all: error
 
929
                    all_versioned = False
 
930
                    break
 
931
                found_versioned = False
 
932
                # for each id at this path
 
933
                for entry in path_entries:
 
934
                    # for each tree.
 
935
                    for index in search_indexes:
 
936
                        if entry[1][index][0] != 'a': # absent
 
937
                            found_versioned = True
 
938
                            # all good: found a versioned cell
 
939
                            break
 
940
                if not found_versioned:
 
941
                    # none of the indexes was not 'absent' at all ids for this
 
942
                    # path.
 
943
                    all_versioned = False
 
944
                    break
 
945
            if not all_versioned:
 
946
                raise errors.PathsNotVersionedError(paths)
 
947
        # -- remove redundancy in supplied paths to prevent over-scanning --
 
948
        search_paths = osutils.minimum_path_selection(paths)
 
949
        # sketch: 
 
950
        # for all search_indexs in each path at or under each element of
 
951
        # search_paths, if the detail is relocated: add the id, and add the
 
952
        # relocated path as one to search if its not searched already. If the
 
953
        # detail is not relocated, add the id.
 
954
        searched_paths = set()
 
955
        found_ids = set()
 
956
        def _process_entry(entry):
 
957
            """Look at search_indexes within entry.
 
958
 
 
959
            If a specific tree's details are relocated, add the relocation
 
960
            target to search_paths if not searched already. If it is absent, do
 
961
            nothing. Otherwise add the id to found_ids.
 
962
            """
 
963
            for index in search_indexes:
 
964
                if entry[1][index][0] == 'r': # relocated
 
965
                    if not osutils.is_inside_any(searched_paths, entry[1][index][1]):
 
966
                        search_paths.add(entry[1][index][1])
 
967
                elif entry[1][index][0] != 'a': # absent
 
968
                    found_ids.add(entry[0][2])
 
969
        while search_paths:
 
970
            current_root = search_paths.pop()
 
971
            searched_paths.add(current_root)
 
972
            # process the entries for this containing directory: the rest will be
 
973
            # found by their parents recursively.
 
974
            root_entries = _entries_for_path(current_root)
 
975
            if not root_entries:
 
976
                # this specified path is not present at all, skip it.
 
977
                continue
 
978
            for entry in root_entries:
 
979
                _process_entry(entry)
 
980
            initial_key = (current_root, '', '')
 
981
            block_index, _ = state._find_block_index_from_key(initial_key)
 
982
            while (block_index < len(state._dirblocks) and
 
983
                osutils.is_inside(current_root, state._dirblocks[block_index][0])):
 
984
                for entry in state._dirblocks[block_index][1]:
 
985
                    _process_entry(entry)
 
986
                block_index += 1
 
987
        return found_ids
 
988
 
 
989
    def _paths2ids_using_bisect(self, paths, search_indexes,
 
990
                                require_versioned=True):
 
991
        state = self.current_dirstate()
 
992
        found_ids = set()
 
993
 
 
994
        split_paths = sorted(osutils.split(p) for p in paths)
 
995
        found = state._bisect_recursive(split_paths)
 
996
 
 
997
        if require_versioned:
 
998
            found_dir_names = set(dir_name_id[:2] for dir_name_id in found)
 
999
            for dir_name in split_paths:
 
1000
                if dir_name not in found_dir_names:
 
1001
                    raise errors.PathsNotVersionedError(paths)
 
1002
 
 
1003
        for dir_name_id, trees_info in found.iteritems():
 
1004
            for index in search_indexes:
 
1005
                if trees_info[index][0] not in ('r', 'a'):
 
1006
                    found_ids.add(dir_name_id[2])
 
1007
        return found_ids
 
1008
 
 
1009
    def read_working_inventory(self):
 
1010
        """Read the working inventory.
 
1011
        
 
1012
        This is a meaningless operation for dirstate, but we obey it anyhow.
 
1013
        """
 
1014
        return self.inventory
 
1015
 
 
1016
    @needs_read_lock
 
1017
    def revision_tree(self, revision_id):
 
1018
        """See Tree.revision_tree.
 
1019
 
 
1020
        WorkingTree4 supplies revision_trees for any basis tree.
 
1021
        """
 
1022
        dirstate = self.current_dirstate()
 
1023
        parent_ids = dirstate.get_parent_ids()
 
1024
        if revision_id not in parent_ids:
 
1025
            raise errors.NoSuchRevisionInTree(self, revision_id)
 
1026
        if revision_id in dirstate.get_ghosts():
 
1027
            raise errors.NoSuchRevisionInTree(self, revision_id)
 
1028
        return DirStateRevisionTree(dirstate, revision_id,
 
1029
            self.branch.repository)
 
1030
 
 
1031
    @needs_tree_write_lock
 
1032
    def set_last_revision(self, new_revision):
 
1033
        """Change the last revision in the working tree."""
 
1034
        parents = self.get_parent_ids()
 
1035
        if new_revision in (NULL_REVISION, None):
 
1036
            assert len(parents) < 2, (
 
1037
                "setting the last parent to none with a pending merge is "
 
1038
                "unsupported.")
 
1039
            self.set_parent_ids([])
 
1040
        else:
 
1041
            self.set_parent_ids([new_revision] + parents[1:],
 
1042
                allow_leftmost_as_ghost=True)
 
1043
 
 
1044
    @needs_tree_write_lock
 
1045
    def set_parent_ids(self, revision_ids, allow_leftmost_as_ghost=False):
 
1046
        """Set the parent ids to revision_ids.
 
1047
        
 
1048
        See also set_parent_trees. This api will try to retrieve the tree data
 
1049
        for each element of revision_ids from the trees repository. If you have
 
1050
        tree data already available, it is more efficient to use
 
1051
        set_parent_trees rather than set_parent_ids. set_parent_ids is however
 
1052
        an easier API to use.
 
1053
 
 
1054
        :param revision_ids: The revision_ids to set as the parent ids of this
 
1055
            working tree. Any of these may be ghosts.
 
1056
        """
 
1057
        trees = []
 
1058
        for revision_id in revision_ids:
 
1059
            try:
 
1060
                revtree = self.branch.repository.revision_tree(revision_id)
 
1061
                # TODO: jam 20070213 KnitVersionedFile raises
 
1062
                #       RevisionNotPresent rather than NoSuchRevision if a
 
1063
                #       given revision_id is not present. Should Repository be
 
1064
                #       catching it and re-raising NoSuchRevision?
 
1065
            except (errors.NoSuchRevision, errors.RevisionNotPresent):
 
1066
                revtree = None
 
1067
            trees.append((revision_id, revtree))
 
1068
        self.set_parent_trees(trees,
 
1069
            allow_leftmost_as_ghost=allow_leftmost_as_ghost)
 
1070
 
 
1071
    @needs_tree_write_lock
 
1072
    def set_parent_trees(self, parents_list, allow_leftmost_as_ghost=False):
 
1073
        """Set the parents of the working tree.
 
1074
 
 
1075
        :param parents_list: A list of (revision_id, tree) tuples.
 
1076
            If tree is None, then that element is treated as an unreachable
 
1077
            parent tree - i.e. a ghost.
 
1078
        """
 
1079
        dirstate = self.current_dirstate()
 
1080
        if len(parents_list) > 0:
 
1081
            if not allow_leftmost_as_ghost and parents_list[0][1] is None:
 
1082
                raise errors.GhostRevisionUnusableHere(parents_list[0][0])
 
1083
        real_trees = []
 
1084
        ghosts = []
 
1085
        # convert absent trees to the null tree, which we convert back to
 
1086
        # missing on access.
 
1087
        for rev_id, tree in parents_list:
 
1088
            _mod_revision.check_not_reserved_id(rev_id)
 
1089
            if tree is not None:
 
1090
                real_trees.append((rev_id, tree))
 
1091
            else:
 
1092
                real_trees.append((rev_id,
 
1093
                    self.branch.repository.revision_tree(None)))
 
1094
                ghosts.append(rev_id)
 
1095
        dirstate.set_parent_trees(real_trees, ghosts=ghosts)
 
1096
        self._make_dirty(reset_inventory=False)
 
1097
 
 
1098
    def _set_root_id(self, file_id):
 
1099
        """See WorkingTree.set_root_id."""
 
1100
        state = self.current_dirstate()
 
1101
        state.set_path_id('', file_id)
 
1102
        if state._dirblock_state == dirstate.DirState.IN_MEMORY_MODIFIED:
 
1103
            self._make_dirty(reset_inventory=True)
 
1104
 
 
1105
    def _sha_from_stat(self, path, stat_result):
 
1106
        """Get a sha digest from the tree's stat cache.
 
1107
 
 
1108
        The default implementation assumes no stat cache is present.
 
1109
 
 
1110
        :param path: The path.
 
1111
        :param stat_result: The stat result being looked up.
 
1112
        """
 
1113
        return self.current_dirstate().sha1_from_stat(path, stat_result)
 
1114
 
 
1115
    @needs_read_lock
 
1116
    def supports_tree_reference(self):
 
1117
        return self._repo_supports_tree_reference
 
1118
 
 
1119
    def unlock(self):
 
1120
        """Unlock in format 4 trees needs to write the entire dirstate."""
 
1121
        # do non-implementation specific cleanup
 
1122
        self._cleanup()
 
1123
 
 
1124
        if self._control_files._lock_count == 1:
 
1125
            # eventually we should do signature checking during read locks for
 
1126
            # dirstate updates.
 
1127
            if self._control_files._lock_mode == 'w':
 
1128
                if self._dirty:
 
1129
                    self.flush()
 
1130
            if self._dirstate is not None:
 
1131
                # This is a no-op if there are no modifications.
 
1132
                self._dirstate.save()
 
1133
                self._dirstate.unlock()
 
1134
            # TODO: jam 20070301 We shouldn't have to wipe the dirstate at this
 
1135
            #       point. Instead, it could check if the header has been
 
1136
            #       modified when it is locked, and if not, it can hang on to
 
1137
            #       the data it has in memory.
 
1138
            self._dirstate = None
 
1139
            self._inventory = None
 
1140
        # reverse order of locking.
 
1141
        try:
 
1142
            return self._control_files.unlock()
 
1143
        finally:
 
1144
            self.branch.unlock()
 
1145
 
 
1146
    @needs_tree_write_lock
 
1147
    def unversion(self, file_ids):
 
1148
        """Remove the file ids in file_ids from the current versioned set.
 
1149
 
 
1150
        When a file_id is unversioned, all of its children are automatically
 
1151
        unversioned.
 
1152
 
 
1153
        :param file_ids: The file ids to stop versioning.
 
1154
        :raises: NoSuchId if any fileid is not currently versioned.
 
1155
        """
 
1156
        if not file_ids:
 
1157
            return
 
1158
        state = self.current_dirstate()
 
1159
        state._read_dirblocks_if_needed()
 
1160
        ids_to_unversion = set(file_ids)
 
1161
        paths_to_unversion = set()
 
1162
        # sketch:
 
1163
        # check if the root is to be unversioned, if so, assert for now.
 
1164
        # walk the state marking unversioned things as absent.
 
1165
        # if there are any un-unversioned ids at the end, raise
 
1166
        for key, details in state._dirblocks[0][1]:
 
1167
            if (details[0][0] not in ('a', 'r') and # absent or relocated
 
1168
                key[2] in ids_to_unversion):
 
1169
                # I haven't written the code to unversion / yet - it should be
 
1170
                # supported.
 
1171
                raise errors.BzrError('Unversioning the / is not currently supported')
 
1172
        block_index = 0
 
1173
        while block_index < len(state._dirblocks):
 
1174
            # process one directory at a time.
 
1175
            block = state._dirblocks[block_index]
 
1176
            # first check: is the path one to remove - it or its children
 
1177
            delete_block = False
 
1178
            for path in paths_to_unversion:
 
1179
                if (block[0].startswith(path) and
 
1180
                    (len(block[0]) == len(path) or
 
1181
                     block[0][len(path)] == '/')):
 
1182
                    # this entire block should be deleted - its the block for a
 
1183
                    # path to unversion; or the child of one
 
1184
                    delete_block = True
 
1185
                    break
 
1186
            # TODO: trim paths_to_unversion as we pass by paths
 
1187
            if delete_block:
 
1188
                # this block is to be deleted: process it.
 
1189
                # TODO: we can special case the no-parents case and
 
1190
                # just forget the whole block.
 
1191
                entry_index = 0
 
1192
                while entry_index < len(block[1]):
 
1193
                    # Mark this file id as having been removed
 
1194
                    entry = block[1][entry_index]
 
1195
                    ids_to_unversion.discard(entry[0][2])
 
1196
                    if (entry[1][0][0] in 'ar' # don't remove absent or renamed
 
1197
                                               # entries
 
1198
                        or not state._make_absent(entry)):
 
1199
                        entry_index += 1
 
1200
                # go to the next block. (At the moment we dont delete empty
 
1201
                # dirblocks)
 
1202
                block_index += 1
 
1203
                continue
 
1204
            entry_index = 0
 
1205
            while entry_index < len(block[1]):
 
1206
                entry = block[1][entry_index]
 
1207
                if (entry[1][0][0] in ('a', 'r') or # absent, relocated
 
1208
                    # ^ some parent row.
 
1209
                    entry[0][2] not in ids_to_unversion):
 
1210
                    # ^ not an id to unversion
 
1211
                    entry_index += 1
 
1212
                    continue
 
1213
                if entry[1][0][0] == 'd':
 
1214
                    paths_to_unversion.add(pathjoin(entry[0][0], entry[0][1]))
 
1215
                if not state._make_absent(entry):
 
1216
                    entry_index += 1
 
1217
                # we have unversioned this id
 
1218
                ids_to_unversion.remove(entry[0][2])
 
1219
            block_index += 1
 
1220
        if ids_to_unversion:
 
1221
            raise errors.NoSuchId(self, iter(ids_to_unversion).next())
 
1222
        self._make_dirty(reset_inventory=False)
 
1223
        # have to change the legacy inventory too.
 
1224
        if self._inventory is not None:
 
1225
            for file_id in file_ids:
 
1226
                self._inventory.remove_recursive_id(file_id)
 
1227
 
 
1228
    @needs_tree_write_lock
 
1229
    def rename_one(self, from_rel, to_rel, after=False):
 
1230
        """See WorkingTree.rename_one"""
 
1231
        self.flush()
 
1232
        WorkingTree.rename_one(self, from_rel, to_rel, after)
 
1233
 
 
1234
    @needs_tree_write_lock
 
1235
    def apply_inventory_delta(self, changes):
 
1236
        """See MutableTree.apply_inventory_delta"""
 
1237
        state = self.current_dirstate()
 
1238
        state.update_by_delta(changes)
 
1239
        self._make_dirty(reset_inventory=True)
 
1240
 
 
1241
    def update_basis_by_delta(self, new_revid, delta):
 
1242
        """See MutableTree.update_basis_by_delta."""
 
1243
        assert self.last_revision() != new_revid
 
1244
        self.current_dirstate().update_basis_by_delta(delta, new_revid)
 
1245
 
 
1246
    @needs_read_lock
 
1247
    def _validate(self):
 
1248
        self._dirstate._validate()
 
1249
 
 
1250
    @needs_tree_write_lock
 
1251
    def _write_inventory(self, inv):
 
1252
        """Write inventory as the current inventory."""
 
1253
        assert not self._dirty, ("attempting to write an inventory when the "
 
1254
            "dirstate is dirty will cause data loss")
 
1255
        self.current_dirstate().set_state_from_inventory(inv)
 
1256
        self._make_dirty(reset_inventory=False)
 
1257
        if self._inventory is not None:
 
1258
            self._inventory = inv
 
1259
        self.flush()
 
1260
 
 
1261
 
 
1262
class WorkingTreeFormat4(WorkingTreeFormat3):
 
1263
    """The first consolidated dirstate working tree format.
 
1264
 
 
1265
    This format:
 
1266
        - exists within a metadir controlling .bzr
 
1267
        - includes an explicit version marker for the workingtree control
 
1268
          files, separate from the BzrDir format
 
1269
        - modifies the hash cache format
 
1270
        - is new in bzr 0.15
 
1271
        - uses a LockDir to guard access to it.
 
1272
    """
 
1273
 
 
1274
    upgrade_recommended = False
 
1275
 
 
1276
    def get_format_string(self):
 
1277
        """See WorkingTreeFormat.get_format_string()."""
 
1278
        return "Bazaar Working Tree Format 4 (bzr 0.15)\n"
 
1279
 
 
1280
    def get_format_description(self):
 
1281
        """See WorkingTreeFormat.get_format_description()."""
 
1282
        return "Working tree format 4"
 
1283
 
 
1284
    def initialize(self, a_bzrdir, revision_id=None, from_branch=None,
 
1285
                   accelerator_tree=None, hardlink=False):
 
1286
        """See WorkingTreeFormat.initialize().
 
1287
 
 
1288
        :param revision_id: allows creating a working tree at a different
 
1289
        revision than the branch is at.
 
1290
        :param accelerator_tree: A tree which can be used for retrieving file
 
1291
            contents more quickly than the revision tree, i.e. a workingtree.
 
1292
            The revision tree will be used for cases where accelerator_tree's
 
1293
            content is different.
 
1294
        :param hardlink: If true, hard-link files from accelerator_tree,
 
1295
            where possible.
 
1296
 
 
1297
        These trees get an initial random root id, if their repository supports
 
1298
        rich root data, TREE_ROOT otherwise.
 
1299
        """
 
1300
        if not isinstance(a_bzrdir.transport, LocalTransport):
 
1301
            raise errors.NotLocalUrl(a_bzrdir.transport.base)
 
1302
        transport = a_bzrdir.get_workingtree_transport(self)
 
1303
        control_files = self._open_control_files(a_bzrdir)
 
1304
        control_files.create_lock()
 
1305
        control_files.lock_write()
 
1306
        control_files.put_utf8('format', self.get_format_string())
 
1307
        if from_branch is not None:
 
1308
            branch = from_branch
 
1309
        else:
 
1310
            branch = a_bzrdir.open_branch()
 
1311
        if revision_id is None:
 
1312
            revision_id = branch.last_revision()
 
1313
        local_path = transport.local_abspath('dirstate')
 
1314
        # write out new dirstate (must exist when we create the tree)
 
1315
        state = dirstate.DirState.initialize(local_path)
 
1316
        state.unlock()
 
1317
        del state
 
1318
        wt = WorkingTree4(a_bzrdir.root_transport.local_abspath('.'),
 
1319
                         branch,
 
1320
                         _format=self,
 
1321
                         _bzrdir=a_bzrdir,
 
1322
                         _control_files=control_files)
 
1323
        wt._new_tree()
 
1324
        wt.lock_tree_write()
 
1325
        try:
 
1326
            if revision_id in (None, NULL_REVISION):
 
1327
                if branch.repository.supports_rich_root():
 
1328
                    wt._set_root_id(generate_ids.gen_root_id())
 
1329
                else:
 
1330
                    wt._set_root_id(ROOT_ID)
 
1331
                wt.flush()
 
1332
            basis = None
 
1333
            # frequently, we will get here due to branching.  The accelerator
 
1334
            # tree will be the tree from the branch, so the desired basis
 
1335
            # tree will often be a parent of the accelerator tree.
 
1336
            if accelerator_tree is not None:
 
1337
                try:
 
1338
                    basis = accelerator_tree.revision_tree(revision_id)
 
1339
                except errors.NoSuchRevision:
 
1340
                    pass
 
1341
            if basis is None:
 
1342
                basis = branch.repository.revision_tree(revision_id)
 
1343
            if revision_id == NULL_REVISION:
 
1344
                parents_list = []
 
1345
            else:
 
1346
                parents_list = [(revision_id, basis)]
 
1347
            basis.lock_read()
 
1348
            try:
 
1349
                wt.set_parent_trees(parents_list, allow_leftmost_as_ghost=True)
 
1350
                wt.flush()
 
1351
                # if the basis has a root id we have to use that; otherwise we
 
1352
                # use a new random one
 
1353
                basis_root_id = basis.get_root_id()
 
1354
                if basis_root_id is not None:
 
1355
                    wt._set_root_id(basis_root_id)
 
1356
                    wt.flush()
 
1357
                transform.build_tree(basis, wt, accelerator_tree,
 
1358
                                     hardlink=hardlink)
 
1359
            finally:
 
1360
                basis.unlock()
 
1361
        finally:
 
1362
            control_files.unlock()
 
1363
            wt.unlock()
 
1364
        return wt
 
1365
 
 
1366
    def _open(self, a_bzrdir, control_files):
 
1367
        """Open the tree itself.
 
1368
 
 
1369
        :param a_bzrdir: the dir for the tree.
 
1370
        :param control_files: the control files for the tree.
 
1371
        """
 
1372
        return WorkingTree4(a_bzrdir.root_transport.local_abspath('.'),
 
1373
                           branch=a_bzrdir.open_branch(),
 
1374
                           _format=self,
 
1375
                           _bzrdir=a_bzrdir,
 
1376
                           _control_files=control_files)
 
1377
 
 
1378
    def __get_matchingbzrdir(self):
 
1379
        # please test against something that will let us do tree references
 
1380
        return bzrdir.format_registry.make_bzrdir(
 
1381
            'dirstate-with-subtree')
 
1382
 
 
1383
    _matchingbzrdir = property(__get_matchingbzrdir)
 
1384
 
 
1385
 
 
1386
class DirStateRevisionTree(Tree):
 
1387
    """A revision tree pulling the inventory from a dirstate."""
 
1388
 
 
1389
    def __init__(self, dirstate, revision_id, repository):
 
1390
        self._dirstate = dirstate
 
1391
        self._revision_id = revision_id
 
1392
        self._repository = repository
 
1393
        self._inventory = None
 
1394
        self._locked = 0
 
1395
        self._dirstate_locked = False
 
1396
 
 
1397
    def __repr__(self):
 
1398
        return "<%s of %s in %s>" % \
 
1399
            (self.__class__.__name__, self._revision_id, self._dirstate)
 
1400
 
 
1401
    def annotate_iter(self, file_id,
 
1402
                      default_revision=_mod_revision.CURRENT_REVISION):
 
1403
        """See Tree.annotate_iter"""
 
1404
        w = self._get_weave(file_id)
 
1405
        return w.annotate_iter(self.inventory[file_id].revision)
 
1406
 
 
1407
    def _get_ancestors(self, default_revision):
 
1408
        return set(self._repository.get_ancestry(self._revision_id,
 
1409
                                                 topo_sorted=False))
 
1410
    def _comparison_data(self, entry, path):
 
1411
        """See Tree._comparison_data."""
 
1412
        if entry is None:
 
1413
            return None, False, None
 
1414
        # trust the entry as RevisionTree does, but this may not be
 
1415
        # sensible: the entry might not have come from us?
 
1416
        return entry.kind, entry.executable, None
 
1417
 
 
1418
    def _file_size(self, entry, stat_value):
 
1419
        return entry.text_size
 
1420
 
 
1421
    def filter_unversioned_files(self, paths):
 
1422
        """Filter out paths that are not versioned.
 
1423
 
 
1424
        :return: set of paths.
 
1425
        """
 
1426
        pred = self.has_filename
 
1427
        return set((p for p in paths if not pred(p)))
 
1428
 
 
1429
    def get_root_id(self):
 
1430
        return self.path2id('')
 
1431
 
 
1432
    def id2path(self, file_id):
 
1433
        "Convert a file-id to a path."
 
1434
        entry = self._get_entry(file_id=file_id)
 
1435
        if entry == (None, None):
 
1436
            raise errors.NoSuchId(tree=self, file_id=file_id)
 
1437
        path_utf8 = osutils.pathjoin(entry[0][0], entry[0][1])
 
1438
        return path_utf8.decode('utf8')
 
1439
 
 
1440
    def _get_parent_index(self):
 
1441
        """Return the index in the dirstate referenced by this tree."""
 
1442
        return self._dirstate.get_parent_ids().index(self._revision_id) + 1
 
1443
 
 
1444
    def _get_entry(self, file_id=None, path=None):
 
1445
        """Get the dirstate row for file_id or path.
 
1446
 
 
1447
        If either file_id or path is supplied, it is used as the key to lookup.
 
1448
        If both are supplied, the fastest lookup is used, and an error is
 
1449
        raised if they do not both point at the same row.
 
1450
        
 
1451
        :param file_id: An optional unicode file_id to be looked up.
 
1452
        :param path: An optional unicode path to be looked up.
 
1453
        :return: The dirstate row tuple for path/file_id, or (None, None)
 
1454
        """
 
1455
        if file_id is None and path is None:
 
1456
            raise errors.BzrError('must supply file_id or path')
 
1457
        if path is not None:
 
1458
            path = path.encode('utf8')
 
1459
        parent_index = self._get_parent_index()
 
1460
        return self._dirstate._get_entry(parent_index, fileid_utf8=file_id, path_utf8=path)
 
1461
 
 
1462
    def _generate_inventory(self):
 
1463
        """Create and set self.inventory from the dirstate object.
 
1464
 
 
1465
        (So this is only called the first time the inventory is requested for
 
1466
        this tree; it then remains in memory until it's out of date.)
 
1467
 
 
1468
        This is relatively expensive: we have to walk the entire dirstate.
 
1469
        """
 
1470
        assert self._locked, 'cannot generate inventory of an unlocked '\
 
1471
            'dirstate revision tree'
 
1472
        # separate call for profiling - makes it clear where the costs are.
 
1473
        self._dirstate._read_dirblocks_if_needed()
 
1474
        assert self._revision_id in self._dirstate.get_parent_ids(), \
 
1475
            'parent %s has disappeared from %s' % (
 
1476
            self._revision_id, self._dirstate.get_parent_ids())
 
1477
        parent_index = self._dirstate.get_parent_ids().index(self._revision_id) + 1
 
1478
        # This is identical now to the WorkingTree _generate_inventory except
 
1479
        # for the tree index use.
 
1480
        root_key, current_entry = self._dirstate._get_entry(parent_index, path_utf8='')
 
1481
        current_id = root_key[2]
 
1482
        assert current_entry[parent_index][0] == 'd'
 
1483
        inv = Inventory(root_id=current_id, revision_id=self._revision_id)
 
1484
        inv.root.revision = current_entry[parent_index][4]
 
1485
        # Turn some things into local variables
 
1486
        minikind_to_kind = dirstate.DirState._minikind_to_kind
 
1487
        factory = entry_factory
 
1488
        utf8_decode = cache_utf8._utf8_decode
 
1489
        inv_byid = inv._byid
 
1490
        # we could do this straight out of the dirstate; it might be fast
 
1491
        # and should be profiled - RBC 20070216
 
1492
        parent_ies = {'' : inv.root}
 
1493
        for block in self._dirstate._dirblocks[1:]: #skip root
 
1494
            dirname = block[0]
 
1495
            try:
 
1496
                parent_ie = parent_ies[dirname]
 
1497
            except KeyError:
 
1498
                # all the paths in this block are not versioned in this tree
 
1499
                continue
 
1500
            for key, entry in block[1]:
 
1501
                minikind, fingerprint, size, executable, revid = entry[parent_index]
 
1502
                if minikind in ('a', 'r'): # absent, relocated
 
1503
                    # not this tree
 
1504
                    continue
 
1505
                name = key[1]
 
1506
                name_unicode = utf8_decode(name)[0]
 
1507
                file_id = key[2]
 
1508
                kind = minikind_to_kind[minikind]
 
1509
                inv_entry = factory[kind](file_id, name_unicode,
 
1510
                                          parent_ie.file_id)
 
1511
                inv_entry.revision = revid
 
1512
                if kind == 'file':
 
1513
                    inv_entry.executable = executable
 
1514
                    inv_entry.text_size = size
 
1515
                    inv_entry.text_sha1 = fingerprint
 
1516
                elif kind == 'directory':
 
1517
                    parent_ies[(dirname + '/' + name).strip('/')] = inv_entry
 
1518
                elif kind == 'symlink':
 
1519
                    inv_entry.executable = False
 
1520
                    inv_entry.text_size = None
 
1521
                    inv_entry.symlink_target = utf8_decode(fingerprint)[0]
 
1522
                elif kind == 'tree-reference':
 
1523
                    inv_entry.reference_revision = fingerprint or None
 
1524
                else:
 
1525
                    raise AssertionError("cannot convert entry %r into an InventoryEntry"
 
1526
                            % entry)
 
1527
                # These checks cost us around 40ms on a 55k entry tree
 
1528
                assert file_id not in inv_byid
 
1529
                assert name_unicode not in parent_ie.children
 
1530
                inv_byid[file_id] = inv_entry
 
1531
                parent_ie.children[name_unicode] = inv_entry
 
1532
        self._inventory = inv
 
1533
 
 
1534
    def get_file_mtime(self, file_id, path=None):
 
1535
        """Return the modification time for this record.
 
1536
 
 
1537
        We return the timestamp of the last-changed revision.
 
1538
        """
 
1539
        # Make sure the file exists
 
1540
        entry = self._get_entry(file_id, path=path)
 
1541
        if entry == (None, None): # do we raise?
 
1542
            return None
 
1543
        parent_index = self._get_parent_index()
 
1544
        last_changed_revision = entry[1][parent_index][4]
 
1545
        return self._repository.get_revision(last_changed_revision).timestamp
 
1546
 
 
1547
    def get_file_sha1(self, file_id, path=None, stat_value=None):
 
1548
        entry = self._get_entry(file_id=file_id, path=path)
 
1549
        parent_index = self._get_parent_index()
 
1550
        parent_details = entry[1][parent_index]
 
1551
        if parent_details[0] == 'f':
 
1552
            return parent_details[1]
 
1553
        return None
 
1554
 
 
1555
    @symbol_versioning.deprecated_method(symbol_versioning.zero_ninety)
 
1556
    def get_weave(self, file_id):
 
1557
        return self._get_weave(file_id)
 
1558
 
 
1559
    def _get_weave(self, file_id):
 
1560
        return self._repository.weave_store.get_weave(file_id,
 
1561
                self._repository.get_transaction())
 
1562
 
 
1563
    def get_file(self, file_id, path=None):
 
1564
        return StringIO(self.get_file_text(file_id))
 
1565
 
 
1566
    def get_file_lines(self, file_id):
 
1567
        entry = self._get_entry(file_id=file_id)[1]
 
1568
        if entry == None:
 
1569
            raise errors.NoSuchId(tree=self, file_id=file_id)
 
1570
        return self._get_weave(file_id).get_lines(entry[1][4])
 
1571
 
 
1572
    def get_file_size(self, file_id):
 
1573
        return self.inventory[file_id].text_size
 
1574
 
 
1575
    def get_file_text(self, file_id):
 
1576
        return ''.join(self.get_file_lines(file_id))
 
1577
 
 
1578
    def get_reference_revision(self, file_id, path=None):
 
1579
        return self.inventory[file_id].reference_revision
 
1580
 
 
1581
    def iter_files_bytes(self, desired_files):
 
1582
        """See Tree.iter_files_bytes.
 
1583
 
 
1584
        This version is implemented on top of Repository.iter_files_bytes"""
 
1585
        parent_index = self._get_parent_index()
 
1586
        repo_desired_files = []
 
1587
        for file_id, identifier in desired_files:
 
1588
            entry = self._get_entry(file_id)
 
1589
            if entry == (None, None):
 
1590
                raise errors.NoSuchId(self, file_id)
 
1591
            repo_desired_files.append((file_id, entry[1][parent_index][4],
 
1592
                                       identifier))
 
1593
        return self._repository.iter_files_bytes(repo_desired_files)
 
1594
 
 
1595
    def get_symlink_target(self, file_id):
 
1596
        entry = self._get_entry(file_id=file_id)
 
1597
        parent_index = self._get_parent_index()
 
1598
        if entry[1][parent_index][0] != 'l':
 
1599
            return None
 
1600
        else:
 
1601
            # At present, none of the tree implementations supports non-ascii
 
1602
            # symlink targets. So we will just assume that the dirstate path is
 
1603
            # correct.
 
1604
            return entry[1][parent_index][1]
 
1605
 
 
1606
    def get_revision_id(self):
 
1607
        """Return the revision id for this tree."""
 
1608
        return self._revision_id
 
1609
 
 
1610
    def _get_inventory(self):
 
1611
        if self._inventory is not None:
 
1612
            return self._inventory
 
1613
        self._must_be_locked()
 
1614
        self._generate_inventory()
 
1615
        return self._inventory
 
1616
 
 
1617
    inventory = property(_get_inventory,
 
1618
                         doc="Inventory of this Tree")
 
1619
 
 
1620
    def get_parent_ids(self):
 
1621
        """The parents of a tree in the dirstate are not cached."""
 
1622
        return self._repository.get_revision(self._revision_id).parent_ids
 
1623
 
 
1624
    def has_filename(self, filename):
 
1625
        return bool(self.path2id(filename))
 
1626
 
 
1627
    def kind(self, file_id):
 
1628
        entry = self._get_entry(file_id=file_id)[1]
 
1629
        if entry == None:
 
1630
            raise errors.NoSuchId(tree=self, file_id=file_id)
 
1631
        return dirstate.DirState._minikind_to_kind[entry[1][0]]
 
1632
 
 
1633
    def stored_kind(self, file_id):
 
1634
        """See Tree.stored_kind"""
 
1635
        return self.kind(file_id)
 
1636
 
 
1637
    def path_content_summary(self, path):
 
1638
        """See Tree.path_content_summary."""
 
1639
        id = self.inventory.path2id(path)
 
1640
        if id is None:
 
1641
            return ('missing', None, None, None)
 
1642
        entry = self._inventory[id]
 
1643
        kind = entry.kind
 
1644
        if kind == 'file':
 
1645
            return (kind, entry.text_size, entry.executable, entry.text_sha1)
 
1646
        elif kind == 'symlink':
 
1647
            return (kind, None, None, entry.symlink_target)
 
1648
        else:
 
1649
            return (kind, None, None, None)
 
1650
 
 
1651
    def is_executable(self, file_id, path=None):
 
1652
        ie = self.inventory[file_id]
 
1653
        if ie.kind != "file":
 
1654
            return None
 
1655
        return ie.executable
 
1656
 
 
1657
    def list_files(self, include_root=False):
 
1658
        # We use a standard implementation, because DirStateRevisionTree is
 
1659
        # dealing with one of the parents of the current state
 
1660
        inv = self._get_inventory()
 
1661
        entries = inv.iter_entries()
 
1662
        if self.inventory.root is not None and not include_root:
 
1663
            entries.next()
 
1664
        for path, entry in entries:
 
1665
            yield path, 'V', entry.kind, entry.file_id, entry
 
1666
 
 
1667
    def lock_read(self):
 
1668
        """Lock the tree for a set of operations."""
 
1669
        if not self._locked:
 
1670
            self._repository.lock_read()
 
1671
            if self._dirstate._lock_token is None:
 
1672
                self._dirstate.lock_read()
 
1673
                self._dirstate_locked = True
 
1674
        self._locked += 1
 
1675
 
 
1676
    def _must_be_locked(self):
 
1677
        if not self._locked:
 
1678
            raise errors.ObjectNotLocked(self)
 
1679
 
 
1680
    @needs_read_lock
 
1681
    def path2id(self, path):
 
1682
        """Return the id for path in this tree."""
 
1683
        # lookup by path: faster than splitting and walking the ivnentory.
 
1684
        entry = self._get_entry(path=path)
 
1685
        if entry == (None, None):
 
1686
            return None
 
1687
        return entry[0][2]
 
1688
 
 
1689
    def unlock(self):
 
1690
        """Unlock, freeing any cache memory used during the lock."""
 
1691
        # outside of a lock, the inventory is suspect: release it.
 
1692
        self._locked -=1
 
1693
        if not self._locked:
 
1694
            self._inventory = None
 
1695
            self._locked = 0
 
1696
            if self._dirstate_locked:
 
1697
                self._dirstate.unlock()
 
1698
                self._dirstate_locked = False
 
1699
            self._repository.unlock()
 
1700
 
 
1701
    def walkdirs(self, prefix=""):
 
1702
        # TODO: jam 20070215 This is the lazy way by using the RevisionTree
 
1703
        # implementation based on an inventory.  
 
1704
        # This should be cleaned up to use the much faster Dirstate code
 
1705
        # So for now, we just build up the parent inventory, and extract
 
1706
        # it the same way RevisionTree does.
 
1707
        _directory = 'directory'
 
1708
        inv = self._get_inventory()
 
1709
        top_id = inv.path2id(prefix)
 
1710
        if top_id is None:
 
1711
            pending = []
 
1712
        else:
 
1713
            pending = [(prefix, top_id)]
 
1714
        while pending:
 
1715
            dirblock = []
 
1716
            relpath, file_id = pending.pop()
 
1717
            # 0 - relpath, 1- file-id
 
1718
            if relpath:
 
1719
                relroot = relpath + '/'
 
1720
            else:
 
1721
                relroot = ""
 
1722
            # FIXME: stash the node in pending
 
1723
            entry = inv[file_id]
 
1724
            for name, child in entry.sorted_children():
 
1725
                toppath = relroot + name
 
1726
                dirblock.append((toppath, name, child.kind, None,
 
1727
                    child.file_id, child.kind
 
1728
                    ))
 
1729
            yield (relpath, entry.file_id), dirblock
 
1730
            # push the user specified dirs from dirblock
 
1731
            for dir in reversed(dirblock):
 
1732
                if dir[2] == _directory:
 
1733
                    pending.append((dir[0], dir[4]))
 
1734
 
 
1735
 
 
1736
class InterDirStateTree(InterTree):
 
1737
    """Fast path optimiser for changes_from with dirstate trees.
 
1738
    
 
1739
    This is used only when both trees are in the dirstate working file, and 
 
1740
    the source is any parent within the dirstate, and the destination is 
 
1741
    the current working tree of the same dirstate.
 
1742
    """
 
1743
    # this could be generalized to allow comparisons between any trees in the
 
1744
    # dirstate, and possibly between trees stored in different dirstates.
 
1745
 
 
1746
    def __init__(self, source, target):
 
1747
        super(InterDirStateTree, self).__init__(source, target)
 
1748
        if not InterDirStateTree.is_compatible(source, target):
 
1749
            raise Exception, "invalid source %r and target %r" % (source, target)
 
1750
 
 
1751
    @staticmethod
 
1752
    def make_source_parent_tree(source, target):
 
1753
        """Change the source tree into a parent of the target."""
 
1754
        revid = source.commit('record tree')
 
1755
        target.branch.repository.fetch(source.branch.repository, revid)
 
1756
        target.set_parent_ids([revid])
 
1757
        return target.basis_tree(), target
 
1758
 
 
1759
    _matching_from_tree_format = WorkingTreeFormat4()
 
1760
    _matching_to_tree_format = WorkingTreeFormat4()
 
1761
    _test_mutable_trees_to_test_trees = make_source_parent_tree
 
1762
 
 
1763
    def _iter_changes(self, include_unchanged=False,
 
1764
                      specific_files=None, pb=None, extra_trees=[],
 
1765
                      require_versioned=True, want_unversioned=False):
 
1766
        """Return the changes from source to target.
 
1767
 
 
1768
        :return: An iterator that yields tuples. See InterTree._iter_changes
 
1769
            for details.
 
1770
        :param specific_files: An optional list of file paths to restrict the
 
1771
            comparison to. When mapping filenames to ids, all matches in all
 
1772
            trees (including optional extra_trees) are used, and all children of
 
1773
            matched directories are included.
 
1774
        :param include_unchanged: An optional boolean requesting the inclusion of
 
1775
            unchanged entries in the result.
 
1776
        :param extra_trees: An optional list of additional trees to use when
 
1777
            mapping the contents of specific_files (paths) to file_ids.
 
1778
        :param require_versioned: If True, all files in specific_files must be
 
1779
            versioned in one of source, target, extra_trees or
 
1780
            PathsNotVersionedError is raised.
 
1781
        :param want_unversioned: Should unversioned files be returned in the
 
1782
            output. An unversioned file is defined as one with (False, False)
 
1783
            for the versioned pair.
 
1784
        """
 
1785
        utf8_decode = cache_utf8._utf8_decode
 
1786
        _minikind_to_kind = dirstate.DirState._minikind_to_kind
 
1787
        cmp_by_dirs = dirstate.cmp_by_dirs
 
1788
        # NB: show_status depends on being able to pass in non-versioned files
 
1789
        # and report them as unknown
 
1790
        # TODO: handle extra trees in the dirstate.
 
1791
        if (extra_trees or specific_files == []):
 
1792
            # we can't fast-path these cases (yet)
 
1793
            for f in super(InterDirStateTree, self)._iter_changes(
 
1794
                include_unchanged, specific_files, pb, extra_trees,
 
1795
                require_versioned, want_unversioned=want_unversioned):
 
1796
                yield f
 
1797
            return
 
1798
        parent_ids = self.target.get_parent_ids()
 
1799
        assert (self.source._revision_id in parent_ids
 
1800
                or self.source._revision_id == NULL_REVISION), \
 
1801
                "revision {%s} is not stored in {%s}, but %s " \
 
1802
                "can only be used for trees stored in the dirstate" \
 
1803
                % (self.source._revision_id, self.target, self._iter_changes)
 
1804
        target_index = 0
 
1805
        if self.source._revision_id == NULL_REVISION:
 
1806
            source_index = None
 
1807
            indices = (target_index,)
 
1808
        else:
 
1809
            assert (self.source._revision_id in parent_ids), \
 
1810
                "Failure: source._revision_id: %s not in target.parent_ids(%s)" % (
 
1811
                self.source._revision_id, parent_ids)
 
1812
            source_index = 1 + parent_ids.index(self.source._revision_id)
 
1813
            indices = (source_index, target_index)
 
1814
        # -- make all specific_files utf8 --
 
1815
        if specific_files:
 
1816
            specific_files_utf8 = set()
 
1817
            for path in specific_files:
 
1818
                specific_files_utf8.add(path.encode('utf8'))
 
1819
            specific_files = specific_files_utf8
 
1820
        else:
 
1821
            specific_files = set([''])
 
1822
        # -- specific_files is now a utf8 path set --
 
1823
        # -- get the state object and prepare it.
 
1824
        state = self.target.current_dirstate()
 
1825
        state._read_dirblocks_if_needed()
 
1826
        def _entries_for_path(path):
 
1827
            """Return a list with all the entries that match path for all ids.
 
1828
            """
 
1829
            dirname, basename = os.path.split(path)
 
1830
            key = (dirname, basename, '')
 
1831
            block_index, present = state._find_block_index_from_key(key)
 
1832
            if not present:
 
1833
                # the block which should contain path is absent.
 
1834
                return []
 
1835
            result = []
 
1836
            block = state._dirblocks[block_index][1]
 
1837
            entry_index, _ = state._find_entry_index(key, block)
 
1838
            # we may need to look at multiple entries at this path: walk while the specific_files match.
 
1839
            while (entry_index < len(block) and
 
1840
                block[entry_index][0][0:2] == key[0:2]):
 
1841
                result.append(block[entry_index])
 
1842
                entry_index += 1
 
1843
            return result
 
1844
        if require_versioned:
 
1845
            # -- check all supplied paths are versioned in a search tree. --
 
1846
            all_versioned = True
 
1847
            for path in specific_files:
 
1848
                path_entries = _entries_for_path(path)
 
1849
                if not path_entries:
 
1850
                    # this specified path is not present at all: error
 
1851
                    all_versioned = False
 
1852
                    break
 
1853
                found_versioned = False
 
1854
                # for each id at this path
 
1855
                for entry in path_entries:
 
1856
                    # for each tree.
 
1857
                    for index in indices:
 
1858
                        if entry[1][index][0] != 'a': # absent
 
1859
                            found_versioned = True
 
1860
                            # all good: found a versioned cell
 
1861
                            break
 
1862
                if not found_versioned:
 
1863
                    # none of the indexes was not 'absent' at all ids for this
 
1864
                    # path.
 
1865
                    all_versioned = False
 
1866
                    break
 
1867
            if not all_versioned:
 
1868
                raise errors.PathsNotVersionedError(specific_files)
 
1869
        # -- remove redundancy in supplied specific_files to prevent over-scanning --
 
1870
        search_specific_files = set()
 
1871
        for path in specific_files:
 
1872
            other_specific_files = specific_files.difference(set([path]))
 
1873
            if not osutils.is_inside_any(other_specific_files, path):
 
1874
                # this is a top level path, we must check it.
 
1875
                search_specific_files.add(path)
 
1876
        # sketch: 
 
1877
        # compare source_index and target_index at or under each element of search_specific_files.
 
1878
        # follow the following comparison table. Note that we only want to do diff operations when
 
1879
        # the target is fdl because thats when the walkdirs logic will have exposed the pathinfo 
 
1880
        # for the target.
 
1881
        # cases:
 
1882
        # 
 
1883
        # Source | Target | disk | action
 
1884
        #   r    | fdlt   |      | add source to search, add id path move and perform
 
1885
        #        |        |      | diff check on source-target
 
1886
        #   r    | fdlt   |  a   | dangling file that was present in the basis. 
 
1887
        #        |        |      | ???
 
1888
        #   r    |  a     |      | add source to search
 
1889
        #   r    |  a     |  a   | 
 
1890
        #   r    |  r     |      | this path is present in a non-examined tree, skip.
 
1891
        #   r    |  r     |  a   | this path is present in a non-examined tree, skip.
 
1892
        #   a    | fdlt   |      | add new id
 
1893
        #   a    | fdlt   |  a   | dangling locally added file, skip
 
1894
        #   a    |  a     |      | not present in either tree, skip
 
1895
        #   a    |  a     |  a   | not present in any tree, skip
 
1896
        #   a    |  r     |      | not present in either tree at this path, skip as it
 
1897
        #        |        |      | may not be selected by the users list of paths.
 
1898
        #   a    |  r     |  a   | not present in either tree at this path, skip as it
 
1899
        #        |        |      | may not be selected by the users list of paths.
 
1900
        #  fdlt  | fdlt   |      | content in both: diff them
 
1901
        #  fdlt  | fdlt   |  a   | deleted locally, but not unversioned - show as deleted ?
 
1902
        #  fdlt  |  a     |      | unversioned: output deleted id for now
 
1903
        #  fdlt  |  a     |  a   | unversioned and deleted: output deleted id
 
1904
        #  fdlt  |  r     |      | relocated in this tree, so add target to search.
 
1905
        #        |        |      | Dont diff, we will see an r,fd; pair when we reach
 
1906
        #        |        |      | this id at the other path.
 
1907
        #  fdlt  |  r     |  a   | relocated in this tree, so add target to search.
 
1908
        #        |        |      | Dont diff, we will see an r,fd; pair when we reach
 
1909
        #        |        |      | this id at the other path.
 
1910
 
 
1911
        # for all search_indexs in each path at or under each element of
 
1912
        # search_specific_files, if the detail is relocated: add the id, and add the
 
1913
        # relocated path as one to search if its not searched already. If the
 
1914
        # detail is not relocated, add the id.
 
1915
        searched_specific_files = set()
 
1916
        NULL_PARENT_DETAILS = dirstate.DirState.NULL_PARENT_DETAILS
 
1917
        # Using a list so that we can access the values and change them in
 
1918
        # nested scope. Each one is [path, file_id, entry]
 
1919
        last_source_parent = [None, None]
 
1920
        last_target_parent = [None, None]
 
1921
 
 
1922
        use_filesystem_for_exec = (sys.platform != 'win32')
 
1923
 
 
1924
        # Just a sentry, so that _process_entry can say that this
 
1925
        # record is handled, but isn't interesting to process (unchanged)
 
1926
        uninteresting = object()
 
1927
 
 
1928
 
 
1929
        old_dirname_to_file_id = {}
 
1930
        new_dirname_to_file_id = {}
 
1931
        # TODO: jam 20070516 - Avoid the _get_entry lookup overhead by
 
1932
        #       keeping a cache of directories that we have seen.
 
1933
 
 
1934
        def _process_entry(entry, path_info):
 
1935
            """Compare an entry and real disk to generate delta information.
 
1936
 
 
1937
            :param path_info: top_relpath, basename, kind, lstat, abspath for
 
1938
                the path of entry. If None, then the path is considered absent.
 
1939
                (Perhaps we should pass in a concrete entry for this ?)
 
1940
                Basename is returned as a utf8 string because we expect this
 
1941
                tuple will be ignored, and don't want to take the time to
 
1942
                decode.
 
1943
            :return: None if these don't match
 
1944
                     A tuple of information about the change, or
 
1945
                     the object 'uninteresting' if these match, but are
 
1946
                     basically identical.
 
1947
            """
 
1948
            if source_index is None:
 
1949
                source_details = NULL_PARENT_DETAILS
 
1950
            else:
 
1951
                source_details = entry[1][source_index]
 
1952
            target_details = entry[1][target_index]
 
1953
            target_minikind = target_details[0]
 
1954
            if path_info is not None and target_minikind in 'fdlt':
 
1955
                assert target_index == 0
 
1956
                link_or_sha1 = state.update_entry(entry, abspath=path_info[4],
 
1957
                                                  stat_value=path_info[3])
 
1958
                # The entry may have been modified by update_entry
 
1959
                target_details = entry[1][target_index]
 
1960
                target_minikind = target_details[0]
 
1961
            else:
 
1962
                link_or_sha1 = None
 
1963
            file_id = entry[0][2]
 
1964
            source_minikind = source_details[0]
 
1965
            if source_minikind in 'fdltr' and target_minikind in 'fdlt':
 
1966
                # claimed content in both: diff
 
1967
                #   r    | fdlt   |      | add source to search, add id path move and perform
 
1968
                #        |        |      | diff check on source-target
 
1969
                #   r    | fdlt   |  a   | dangling file that was present in the basis.
 
1970
                #        |        |      | ???
 
1971
                if source_minikind in 'r':
 
1972
                    # add the source to the search path to find any children it
 
1973
                    # has.  TODO ? : only add if it is a container ?
 
1974
                    if not osutils.is_inside_any(searched_specific_files,
 
1975
                                                 source_details[1]):
 
1976
                        search_specific_files.add(source_details[1])
 
1977
                    # generate the old path; this is needed for stating later
 
1978
                    # as well.
 
1979
                    old_path = source_details[1]
 
1980
                    old_dirname, old_basename = os.path.split(old_path)
 
1981
                    path = pathjoin(entry[0][0], entry[0][1])
 
1982
                    old_entry = state._get_entry(source_index,
 
1983
                                                 path_utf8=old_path)
 
1984
                    # update the source details variable to be the real
 
1985
                    # location.
 
1986
                    if old_entry == (None, None):
 
1987
                        raise errors.CorruptDirstate(state._filename,
 
1988
                            "entry '%s/%s' is considered renamed from %r"
 
1989
                            " but source does not exist\n"
 
1990
                            "entry: %s" % (entry[0][0], entry[0][1], old_path, entry))
 
1991
                    source_details = old_entry[1][source_index]
 
1992
                    source_minikind = source_details[0]
 
1993
                else:
 
1994
                    old_dirname = entry[0][0]
 
1995
                    old_basename = entry[0][1]
 
1996
                    old_path = path = None
 
1997
                if path_info is None:
 
1998
                    # the file is missing on disk, show as removed.
 
1999
                    content_change = True
 
2000
                    target_kind = None
 
2001
                    target_exec = False
 
2002
                else:
 
2003
                    # source and target are both versioned and disk file is present.
 
2004
                    target_kind = path_info[2]
 
2005
                    if target_kind == 'directory':
 
2006
                        if path is None:
 
2007
                            old_path = path = pathjoin(old_dirname, old_basename)
 
2008
                        new_dirname_to_file_id[path] = file_id
 
2009
                        if source_minikind != 'd':
 
2010
                            content_change = True
 
2011
                        else:
 
2012
                            # directories have no fingerprint
 
2013
                            content_change = False
 
2014
                        target_exec = False
 
2015
                    elif target_kind == 'file':
 
2016
                        if source_minikind != 'f':
 
2017
                            content_change = True
 
2018
                        else:
 
2019
                            # We could check the size, but we already have the
 
2020
                            # sha1 hash.
 
2021
                            content_change = (link_or_sha1 != source_details[1])
 
2022
                        # Target details is updated at update_entry time
 
2023
                        if use_filesystem_for_exec:
 
2024
                            # We don't need S_ISREG here, because we are sure
 
2025
                            # we are dealing with a file.
 
2026
                            target_exec = bool(stat.S_IEXEC & path_info[3].st_mode)
 
2027
                        else:
 
2028
                            target_exec = target_details[3]
 
2029
                    elif target_kind == 'symlink':
 
2030
                        if source_minikind != 'l':
 
2031
                            content_change = True
 
2032
                        else:
 
2033
                            content_change = (link_or_sha1 != source_details[1])
 
2034
                        target_exec = False
 
2035
                    elif target_kind == 'tree-reference':
 
2036
                        if source_minikind != 't':
 
2037
                            content_change = True
 
2038
                        else:
 
2039
                            content_change = False
 
2040
                        target_exec = False
 
2041
                    else:
 
2042
                        raise Exception, "unknown kind %s" % path_info[2]
 
2043
                if source_minikind == 'd':
 
2044
                    if path is None:
 
2045
                        old_path = path = pathjoin(old_dirname, old_basename)
 
2046
                    old_dirname_to_file_id[old_path] = file_id
 
2047
                # parent id is the entry for the path in the target tree
 
2048
                if old_dirname == last_source_parent[0]:
 
2049
                    source_parent_id = last_source_parent[1]
 
2050
                else:
 
2051
                    try:
 
2052
                        source_parent_id = old_dirname_to_file_id[old_dirname]
 
2053
                    except KeyError:
 
2054
                        source_parent_entry = state._get_entry(source_index,
 
2055
                                                               path_utf8=old_dirname)
 
2056
                        source_parent_id = source_parent_entry[0][2]
 
2057
                    if source_parent_id == entry[0][2]:
 
2058
                        # This is the root, so the parent is None
 
2059
                        source_parent_id = None
 
2060
                    else:
 
2061
                        last_source_parent[0] = old_dirname
 
2062
                        last_source_parent[1] = source_parent_id
 
2063
                new_dirname = entry[0][0]
 
2064
                if new_dirname == last_target_parent[0]:
 
2065
                    target_parent_id = last_target_parent[1]
 
2066
                else:
 
2067
                    try:
 
2068
                        target_parent_id = new_dirname_to_file_id[new_dirname]
 
2069
                    except KeyError:
 
2070
                        # TODO: We don't always need to do the lookup, because the
 
2071
                        #       parent entry will be the same as the source entry.
 
2072
                        target_parent_entry = state._get_entry(target_index,
 
2073
                                                               path_utf8=new_dirname)
 
2074
                        assert target_parent_entry != (None, None), (
 
2075
                            "Could not find target parent in wt: %s\nparent of: %s"
 
2076
                            % (new_dirname, entry))
 
2077
                        target_parent_id = target_parent_entry[0][2]
 
2078
                    if target_parent_id == entry[0][2]:
 
2079
                        # This is the root, so the parent is None
 
2080
                        target_parent_id = None
 
2081
                    else:
 
2082
                        last_target_parent[0] = new_dirname
 
2083
                        last_target_parent[1] = target_parent_id
 
2084
 
 
2085
                source_exec = source_details[3]
 
2086
                if (include_unchanged
 
2087
                    or content_change
 
2088
                    or source_parent_id != target_parent_id
 
2089
                    or old_basename != entry[0][1]
 
2090
                    or source_exec != target_exec
 
2091
                    ):
 
2092
                    if old_path is None:
 
2093
                        old_path = path = pathjoin(old_dirname, old_basename)
 
2094
                        old_path_u = utf8_decode(old_path)[0]
 
2095
                        path_u = old_path_u
 
2096
                    else:
 
2097
                        old_path_u = utf8_decode(old_path)[0]
 
2098
                        if old_path == path:
 
2099
                            path_u = old_path_u
 
2100
                        else:
 
2101
                            path_u = utf8_decode(path)[0]
 
2102
                    source_kind = _minikind_to_kind[source_minikind]
 
2103
                    return (entry[0][2],
 
2104
                           (old_path_u, path_u),
 
2105
                           content_change,
 
2106
                           (True, True),
 
2107
                           (source_parent_id, target_parent_id),
 
2108
                           (utf8_decode(old_basename)[0], utf8_decode(entry[0][1])[0]),
 
2109
                           (source_kind, target_kind),
 
2110
                           (source_exec, target_exec))
 
2111
                else:
 
2112
                    return uninteresting
 
2113
            elif source_minikind in 'a' and target_minikind in 'fdlt':
 
2114
                # looks like a new file
 
2115
                if path_info is not None:
 
2116
                    path = pathjoin(entry[0][0], entry[0][1])
 
2117
                    # parent id is the entry for the path in the target tree
 
2118
                    # TODO: these are the same for an entire directory: cache em.
 
2119
                    parent_id = state._get_entry(target_index,
 
2120
                                                 path_utf8=entry[0][0])[0][2]
 
2121
                    if parent_id == entry[0][2]:
 
2122
                        parent_id = None
 
2123
                    if use_filesystem_for_exec:
 
2124
                        # We need S_ISREG here, because we aren't sure if this
 
2125
                        # is a file or not.
 
2126
                        target_exec = bool(
 
2127
                            stat.S_ISREG(path_info[3].st_mode)
 
2128
                            and stat.S_IEXEC & path_info[3].st_mode)
 
2129
                    else:
 
2130
                        target_exec = target_details[3]
 
2131
                    return (entry[0][2],
 
2132
                           (None, utf8_decode(path)[0]),
 
2133
                           True,
 
2134
                           (False, True),
 
2135
                           (None, parent_id),
 
2136
                           (None, utf8_decode(entry[0][1])[0]),
 
2137
                           (None, path_info[2]),
 
2138
                           (None, target_exec))
 
2139
                else:
 
2140
                    # but its not on disk: we deliberately treat this as just
 
2141
                    # never-present. (Why ?! - RBC 20070224)
 
2142
                    pass
 
2143
            elif source_minikind in 'fdlt' and target_minikind in 'a':
 
2144
                # unversioned, possibly, or possibly not deleted: we dont care.
 
2145
                # if its still on disk, *and* theres no other entry at this
 
2146
                # path [we dont know this in this routine at the moment -
 
2147
                # perhaps we should change this - then it would be an unknown.
 
2148
                old_path = pathjoin(entry[0][0], entry[0][1])
 
2149
                # parent id is the entry for the path in the target tree
 
2150
                parent_id = state._get_entry(source_index, path_utf8=entry[0][0])[0][2]
 
2151
                if parent_id == entry[0][2]:
 
2152
                    parent_id = None
 
2153
                return (entry[0][2],
 
2154
                       (utf8_decode(old_path)[0], None),
 
2155
                       True,
 
2156
                       (True, False),
 
2157
                       (parent_id, None),
 
2158
                       (utf8_decode(entry[0][1])[0], None),
 
2159
                       (_minikind_to_kind[source_minikind], None),
 
2160
                       (source_details[3], None))
 
2161
            elif source_minikind in 'fdlt' and target_minikind in 'r':
 
2162
                # a rename; could be a true rename, or a rename inherited from
 
2163
                # a renamed parent. TODO: handle this efficiently. Its not
 
2164
                # common case to rename dirs though, so a correct but slow
 
2165
                # implementation will do.
 
2166
                if not osutils.is_inside_any(searched_specific_files, target_details[1]):
 
2167
                    search_specific_files.add(target_details[1])
 
2168
            elif source_minikind in 'ra' and target_minikind in 'ra':
 
2169
                # neither of the selected trees contain this file,
 
2170
                # so skip over it. This is not currently directly tested, but
 
2171
                # is indirectly via test_too_much.TestCommands.test_conflicts.
 
2172
                pass
 
2173
            else:
 
2174
                raise AssertionError("don't know how to compare "
 
2175
                    "source_minikind=%r, target_minikind=%r"
 
2176
                    % (source_minikind, target_minikind))
 
2177
                ## import pdb;pdb.set_trace()
 
2178
            return None
 
2179
 
 
2180
        while search_specific_files:
 
2181
            # TODO: the pending list should be lexically sorted?  the
 
2182
            # interface doesn't require it.
 
2183
            current_root = search_specific_files.pop()
 
2184
            current_root_unicode = current_root.decode('utf8')
 
2185
            searched_specific_files.add(current_root)
 
2186
            # process the entries for this containing directory: the rest will be
 
2187
            # found by their parents recursively.
 
2188
            root_entries = _entries_for_path(current_root)
 
2189
            root_abspath = self.target.abspath(current_root_unicode)
 
2190
            try:
 
2191
                root_stat = os.lstat(root_abspath)
 
2192
            except OSError, e:
 
2193
                if e.errno == errno.ENOENT:
 
2194
                    # the path does not exist: let _process_entry know that.
 
2195
                    root_dir_info = None
 
2196
                else:
 
2197
                    # some other random error: hand it up.
 
2198
                    raise
 
2199
            else:
 
2200
                root_dir_info = ('', current_root,
 
2201
                    osutils.file_kind_from_stat_mode(root_stat.st_mode), root_stat,
 
2202
                    root_abspath)
 
2203
                if root_dir_info[2] == 'directory':
 
2204
                    if self.target._directory_is_tree_reference(
 
2205
                        current_root.decode('utf8')):
 
2206
                        root_dir_info = root_dir_info[:2] + \
 
2207
                            ('tree-reference',) + root_dir_info[3:]
 
2208
 
 
2209
            if not root_entries and not root_dir_info:
 
2210
                # this specified path is not present at all, skip it.
 
2211
                continue
 
2212
            path_handled = False
 
2213
            for entry in root_entries:
 
2214
                result = _process_entry(entry, root_dir_info)
 
2215
                if result is not None:
 
2216
                    path_handled = True
 
2217
                    if result is not uninteresting:
 
2218
                        yield result
 
2219
            if want_unversioned and not path_handled and root_dir_info:
 
2220
                new_executable = bool(
 
2221
                    stat.S_ISREG(root_dir_info[3].st_mode)
 
2222
                    and stat.S_IEXEC & root_dir_info[3].st_mode)
 
2223
                yield (None,
 
2224
                       (None, current_root_unicode),
 
2225
                       True,
 
2226
                       (False, False),
 
2227
                       (None, None),
 
2228
                       (None, splitpath(current_root_unicode)[-1]),
 
2229
                       (None, root_dir_info[2]),
 
2230
                       (None, new_executable)
 
2231
                      )
 
2232
            initial_key = (current_root, '', '')
 
2233
            block_index, _ = state._find_block_index_from_key(initial_key)
 
2234
            if block_index == 0:
 
2235
                # we have processed the total root already, but because the
 
2236
                # initial key matched it we should skip it here.
 
2237
                block_index +=1
 
2238
            if root_dir_info and root_dir_info[2] == 'tree-reference':
 
2239
                current_dir_info = None
 
2240
            else:
 
2241
                dir_iterator = osutils._walkdirs_utf8(root_abspath, prefix=current_root)
 
2242
                try:
 
2243
                    current_dir_info = dir_iterator.next()
 
2244
                except OSError, e:
 
2245
                    # on win32, python2.4 has e.errno == ERROR_DIRECTORY, but
 
2246
                    # python 2.5 has e.errno == EINVAL,
 
2247
                    #            and e.winerror == ERROR_DIRECTORY
 
2248
                    e_winerror = getattr(e, 'winerror', None)
 
2249
                    win_errors = (ERROR_DIRECTORY, ERROR_PATH_NOT_FOUND)
 
2250
                    # there may be directories in the inventory even though
 
2251
                    # this path is not a file on disk: so mark it as end of
 
2252
                    # iterator
 
2253
                    if e.errno in (errno.ENOENT, errno.ENOTDIR, errno.EINVAL):
 
2254
                        current_dir_info = None
 
2255
                    elif (sys.platform == 'win32'
 
2256
                          and (e.errno in win_errors
 
2257
                               or e_winerror in win_errors)):
 
2258
                        current_dir_info = None
 
2259
                    else:
 
2260
                        raise
 
2261
                else:
 
2262
                    if current_dir_info[0][0] == '':
 
2263
                        # remove .bzr from iteration
 
2264
                        bzr_index = bisect_left(current_dir_info[1], ('.bzr',))
 
2265
                        assert current_dir_info[1][bzr_index][0] == '.bzr'
 
2266
                        del current_dir_info[1][bzr_index]
 
2267
            # walk until both the directory listing and the versioned metadata
 
2268
            # are exhausted. 
 
2269
            if (block_index < len(state._dirblocks) and
 
2270
                osutils.is_inside(current_root, state._dirblocks[block_index][0])):
 
2271
                current_block = state._dirblocks[block_index]
 
2272
            else:
 
2273
                current_block = None
 
2274
            while (current_dir_info is not None or
 
2275
                   current_block is not None):
 
2276
                if (current_dir_info and current_block
 
2277
                    and current_dir_info[0][0] != current_block[0]):
 
2278
                    if cmp_by_dirs(current_dir_info[0][0], current_block[0]) < 0:
 
2279
                        # filesystem data refers to paths not covered by the dirblock.
 
2280
                        # this has two possibilities:
 
2281
                        # A) it is versioned but empty, so there is no block for it
 
2282
                        # B) it is not versioned.
 
2283
 
 
2284
                        # if (A) then we need to recurse into it to check for
 
2285
                        # new unknown files or directories.
 
2286
                        # if (B) then we should ignore it, because we don't
 
2287
                        # recurse into unknown directories.
 
2288
                        path_index = 0
 
2289
                        while path_index < len(current_dir_info[1]):
 
2290
                                current_path_info = current_dir_info[1][path_index]
 
2291
                                if want_unversioned:
 
2292
                                    if current_path_info[2] == 'directory':
 
2293
                                        if self.target._directory_is_tree_reference(
 
2294
                                            current_path_info[0].decode('utf8')):
 
2295
                                            current_path_info = current_path_info[:2] + \
 
2296
                                                ('tree-reference',) + current_path_info[3:]
 
2297
                                    new_executable = bool(
 
2298
                                        stat.S_ISREG(current_path_info[3].st_mode)
 
2299
                                        and stat.S_IEXEC & current_path_info[3].st_mode)
 
2300
                                    yield (None,
 
2301
                                        (None, utf8_decode(current_path_info[0])[0]),
 
2302
                                        True,
 
2303
                                        (False, False),
 
2304
                                        (None, None),
 
2305
                                        (None, utf8_decode(current_path_info[1])[0]),
 
2306
                                        (None, current_path_info[2]),
 
2307
                                        (None, new_executable))
 
2308
                                # dont descend into this unversioned path if it is
 
2309
                                # a dir
 
2310
                                if current_path_info[2] in ('directory',
 
2311
                                                            'tree-reference'):
 
2312
                                    del current_dir_info[1][path_index]
 
2313
                                    path_index -= 1
 
2314
                                path_index += 1
 
2315
 
 
2316
                        # This dir info has been handled, go to the next
 
2317
                        try:
 
2318
                            current_dir_info = dir_iterator.next()
 
2319
                        except StopIteration:
 
2320
                            current_dir_info = None
 
2321
                    else:
 
2322
                        # We have a dirblock entry for this location, but there
 
2323
                        # is no filesystem path for this. This is most likely
 
2324
                        # because a directory was removed from the disk.
 
2325
                        # We don't have to report the missing directory,
 
2326
                        # because that should have already been handled, but we
 
2327
                        # need to handle all of the files that are contained
 
2328
                        # within.
 
2329
                        for current_entry in current_block[1]:
 
2330
                            # entry referring to file not present on disk.
 
2331
                            # advance the entry only, after processing.
 
2332
                            result = _process_entry(current_entry, None)
 
2333
                            if result is not None:
 
2334
                                if result is not uninteresting:
 
2335
                                    yield result
 
2336
                        block_index +=1
 
2337
                        if (block_index < len(state._dirblocks) and
 
2338
                            osutils.is_inside(current_root,
 
2339
                                              state._dirblocks[block_index][0])):
 
2340
                            current_block = state._dirblocks[block_index]
 
2341
                        else:
 
2342
                            current_block = None
 
2343
                    continue
 
2344
                entry_index = 0
 
2345
                if current_block and entry_index < len(current_block[1]):
 
2346
                    current_entry = current_block[1][entry_index]
 
2347
                else:
 
2348
                    current_entry = None
 
2349
                advance_entry = True
 
2350
                path_index = 0
 
2351
                if current_dir_info and path_index < len(current_dir_info[1]):
 
2352
                    current_path_info = current_dir_info[1][path_index]
 
2353
                    if current_path_info[2] == 'directory':
 
2354
                        if self.target._directory_is_tree_reference(
 
2355
                            current_path_info[0].decode('utf8')):
 
2356
                            current_path_info = current_path_info[:2] + \
 
2357
                                ('tree-reference',) + current_path_info[3:]
 
2358
                else:
 
2359
                    current_path_info = None
 
2360
                advance_path = True
 
2361
                path_handled = False
 
2362
                while (current_entry is not None or
 
2363
                    current_path_info is not None):
 
2364
                    if current_entry is None:
 
2365
                        # the check for path_handled when the path is adnvaced
 
2366
                        # will yield this path if needed.
 
2367
                        pass
 
2368
                    elif current_path_info is None:
 
2369
                        # no path is fine: the per entry code will handle it.
 
2370
                        result = _process_entry(current_entry, current_path_info)
 
2371
                        if result is not None:
 
2372
                            if result is not uninteresting:
 
2373
                                yield result
 
2374
                    elif (current_entry[0][1] != current_path_info[1]
 
2375
                          or current_entry[1][target_index][0] in 'ar'):
 
2376
                        # The current path on disk doesn't match the dirblock
 
2377
                        # record. Either the dirblock is marked as absent, or
 
2378
                        # the file on disk is not present at all in the
 
2379
                        # dirblock. Either way, report about the dirblock
 
2380
                        # entry, and let other code handle the filesystem one.
 
2381
 
 
2382
                        # Compare the basename for these files to determine
 
2383
                        # which comes first
 
2384
                        if current_path_info[1] < current_entry[0][1]:
 
2385
                            # extra file on disk: pass for now, but only
 
2386
                            # increment the path, not the entry
 
2387
                            advance_entry = False
 
2388
                        else:
 
2389
                            # entry referring to file not present on disk.
 
2390
                            # advance the entry only, after processing.
 
2391
                            result = _process_entry(current_entry, None)
 
2392
                            if result is not None:
 
2393
                                if result is not uninteresting:
 
2394
                                    yield result
 
2395
                            advance_path = False
 
2396
                    else:
 
2397
                        result = _process_entry(current_entry, current_path_info)
 
2398
                        if result is not None:
 
2399
                            path_handled = True
 
2400
                            if result is not uninteresting:
 
2401
                                yield result
 
2402
                    if advance_entry and current_entry is not None:
 
2403
                        entry_index += 1
 
2404
                        if entry_index < len(current_block[1]):
 
2405
                            current_entry = current_block[1][entry_index]
 
2406
                        else:
 
2407
                            current_entry = None
 
2408
                    else:
 
2409
                        advance_entry = True # reset the advance flaga
 
2410
                    if advance_path and current_path_info is not None:
 
2411
                        if not path_handled:
 
2412
                            # unversioned in all regards
 
2413
                            if want_unversioned:
 
2414
                                new_executable = bool(
 
2415
                                    stat.S_ISREG(current_path_info[3].st_mode)
 
2416
                                    and stat.S_IEXEC & current_path_info[3].st_mode)
 
2417
                                yield (None,
 
2418
                                    (None, utf8_decode(current_path_info[0])[0]),
 
2419
                                    True,
 
2420
                                    (False, False),
 
2421
                                    (None, None),
 
2422
                                    (None, utf8_decode(current_path_info[1])[0]),
 
2423
                                    (None, current_path_info[2]),
 
2424
                                    (None, new_executable))
 
2425
                            # dont descend into this unversioned path if it is
 
2426
                            # a dir
 
2427
                            if current_path_info[2] in ('directory'):
 
2428
                                del current_dir_info[1][path_index]
 
2429
                                path_index -= 1
 
2430
                        # dont descend the disk iterator into any tree 
 
2431
                        # paths.
 
2432
                        if current_path_info[2] == 'tree-reference':
 
2433
                            del current_dir_info[1][path_index]
 
2434
                            path_index -= 1
 
2435
                        path_index += 1
 
2436
                        if path_index < len(current_dir_info[1]):
 
2437
                            current_path_info = current_dir_info[1][path_index]
 
2438
                            if current_path_info[2] == 'directory':
 
2439
                                if self.target._directory_is_tree_reference(
 
2440
                                    current_path_info[0].decode('utf8')):
 
2441
                                    current_path_info = current_path_info[:2] + \
 
2442
                                        ('tree-reference',) + current_path_info[3:]
 
2443
                        else:
 
2444
                            current_path_info = None
 
2445
                        path_handled = False
 
2446
                    else:
 
2447
                        advance_path = True # reset the advance flagg.
 
2448
                if current_block is not None:
 
2449
                    block_index += 1
 
2450
                    if (block_index < len(state._dirblocks) and
 
2451
                        osutils.is_inside(current_root, state._dirblocks[block_index][0])):
 
2452
                        current_block = state._dirblocks[block_index]
 
2453
                    else:
 
2454
                        current_block = None
 
2455
                if current_dir_info is not None:
 
2456
                    try:
 
2457
                        current_dir_info = dir_iterator.next()
 
2458
                    except StopIteration:
 
2459
                        current_dir_info = None
 
2460
 
 
2461
 
 
2462
    @staticmethod
 
2463
    def is_compatible(source, target):
 
2464
        # the target must be a dirstate working tree
 
2465
        if not isinstance(target, WorkingTree4):
 
2466
            return False
 
2467
        # the source must be a revtreee or dirstate rev tree.
 
2468
        if not isinstance(source,
 
2469
            (revisiontree.RevisionTree, DirStateRevisionTree)):
 
2470
            return False
 
2471
        # the source revid must be in the target dirstate
 
2472
        if not (source._revision_id == NULL_REVISION or
 
2473
            source._revision_id in target.get_parent_ids()):
 
2474
            # TODO: what about ghosts? it may well need to 
 
2475
            # check for them explicitly.
 
2476
            return False
 
2477
        return True
 
2478
 
 
2479
InterTree.register_optimiser(InterDirStateTree)
 
2480
 
 
2481
 
 
2482
class Converter3to4(object):
 
2483
    """Perform an in-place upgrade of format 3 to format 4 trees."""
 
2484
 
 
2485
    def __init__(self):
 
2486
        self.target_format = WorkingTreeFormat4()
 
2487
 
 
2488
    def convert(self, tree):
 
2489
        # lock the control files not the tree, so that we dont get tree
 
2490
        # on-unlock behaviours, and so that noone else diddles with the 
 
2491
        # tree during upgrade.
 
2492
        tree._control_files.lock_write()
 
2493
        try:
 
2494
            tree.read_working_inventory()
 
2495
            self.create_dirstate_data(tree)
 
2496
            self.update_format(tree)
 
2497
            self.remove_xml_files(tree)
 
2498
        finally:
 
2499
            tree._control_files.unlock()
 
2500
 
 
2501
    def create_dirstate_data(self, tree):
 
2502
        """Create the dirstate based data for tree."""
 
2503
        local_path = tree.bzrdir.get_workingtree_transport(None
 
2504
            ).local_abspath('dirstate')
 
2505
        state = dirstate.DirState.from_tree(tree, local_path)
 
2506
        state.save()
 
2507
        state.unlock()
 
2508
 
 
2509
    def remove_xml_files(self, tree):
 
2510
        """Remove the oldformat 3 data."""
 
2511
        transport = tree.bzrdir.get_workingtree_transport(None)
 
2512
        for path in ['basis-inventory-cache', 'inventory', 'last-revision',
 
2513
            'pending-merges', 'stat-cache']:
 
2514
            try:
 
2515
                transport.delete(path)
 
2516
            except errors.NoSuchFile:
 
2517
                # some files are optional - just deal.
 
2518
                pass
 
2519
 
 
2520
    def update_format(self, tree):
 
2521
        """Change the format marker."""
 
2522
        tree._control_files.put_utf8('format',
 
2523
            self.target_format.get_format_string())