/brz/remove-bazaar

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

« back to all changes in this revision

Viewing changes to bzrlib/workingtree_4.py

  • Committer: Vincent Ladeuil
  • Date: 2007-09-24 11:05:05 UTC
  • mto: (2852.2.1 trunk)
  • mto: This revision was merged to the branch mainline in revision 2853.
  • Revision ID: v.ladeuil+lp@free.fr-20070924110505-4apfu3r909hamrtq
Fix typo.

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
    hashcache,
 
53
    ignores,
 
54
    merge,
 
55
    osutils,
 
56
    revision as _mod_revision,
 
57
    revisiontree,
 
58
    textui,
 
59
    trace,
 
60
    transform,
 
61
    urlutils,
 
62
    xml5,
 
63
    xml6,
 
64
    )
 
65
import bzrlib.branch
 
66
from bzrlib.transport import get_transport
 
67
import bzrlib.ui
 
68
""")
 
69
 
 
70
from bzrlib import symbol_versioning
 
71
from bzrlib.decorators import needs_read_lock, needs_write_lock
 
72
from bzrlib.inventory import InventoryEntry, Inventory, ROOT_ID, entry_factory
 
73
from bzrlib.lockable_files import LockableFiles, TransportLock
 
74
from bzrlib.lockdir import LockDir
 
75
import bzrlib.mutabletree
 
76
from bzrlib.mutabletree import needs_tree_write_lock
 
77
from bzrlib.osutils import (
 
78
    file_kind,
 
79
    isdir,
 
80
    normpath,
 
81
    pathjoin,
 
82
    rand_chars,
 
83
    realpath,
 
84
    safe_unicode,
 
85
    splitpath,
 
86
    )
 
87
from bzrlib.trace import mutter, note
 
88
from bzrlib.transport.local import LocalTransport
 
89
from bzrlib.tree import InterTree
 
90
from bzrlib.progress import DummyProgress, ProgressPhase
 
91
from bzrlib.revision import NULL_REVISION, CURRENT_REVISION
 
92
from bzrlib.rio import RioReader, rio_file, Stanza
 
93
from bzrlib.symbol_versioning import (deprecated_passed,
 
94
        deprecated_method,
 
95
        deprecated_function,
 
96
        DEPRECATED_PARAMETER,
 
97
        )
 
98
from bzrlib.tree import Tree
 
99
from bzrlib.workingtree import WorkingTree, WorkingTree3, WorkingTreeFormat3
 
100
 
 
101
 
 
102
# This is the Windows equivalent of ENOTDIR
 
103
# It is defined in pywin32.winerror, but we don't want a strong dependency for
 
104
# just an error code.
 
105
ERROR_PATH_NOT_FOUND = 3
 
106
ERROR_DIRECTORY = 267
 
107
 
 
108
 
 
109
class WorkingTree4(WorkingTree3):
 
110
    """This is the Format 4 working tree.
 
111
 
 
112
    This differs from WorkingTree3 by:
 
113
     - Having a consolidated internal dirstate, stored in a
 
114
       randomly-accessible sorted file on disk.
 
115
     - Not having a regular inventory attribute.  One can be synthesized 
 
116
       on demand but this is expensive and should be avoided.
 
117
 
 
118
    This is new in bzr 0.15.
 
119
    """
 
120
 
 
121
    def __init__(self, basedir,
 
122
                 branch,
 
123
                 _control_files=None,
 
124
                 _format=None,
 
125
                 _bzrdir=None):
 
126
        """Construct a WorkingTree for basedir.
 
127
 
 
128
        If the branch is not supplied, it is opened automatically.
 
129
        If the branch is supplied, it must be the branch for this basedir.
 
130
        (branch.base is not cross checked, because for remote branches that
 
131
        would be meaningless).
 
132
        """
 
133
        self._format = _format
 
134
        self.bzrdir = _bzrdir
 
135
        assert isinstance(basedir, basestring), \
 
136
            "base directory %r is not a string" % basedir
 
137
        basedir = safe_unicode(basedir)
 
138
        mutter("opening working tree %r", basedir)
 
139
        self._branch = branch
 
140
        assert isinstance(self.branch, bzrlib.branch.Branch), \
 
141
            "branch %r is not a Branch" % self.branch
 
142
        self.basedir = realpath(basedir)
 
143
        # if branch is at our basedir and is a format 6 or less
 
144
        # assume all other formats have their own control files.
 
145
        assert isinstance(_control_files, LockableFiles), \
 
146
            "_control_files must be a LockableFiles, not %r" % _control_files
 
147
        self._control_files = _control_files
 
148
        self._dirty = None
 
149
        #-------------
 
150
        # during a read or write lock these objects are set, and are
 
151
        # None the rest of the time.
 
152
        self._dirstate = None
 
153
        self._inventory = None
 
154
        #-------------
 
155
 
 
156
    @needs_tree_write_lock
 
157
    def _add(self, files, ids, kinds):
 
158
        """See MutableTree._add."""
 
159
        state = self.current_dirstate()
 
160
        for f, file_id, kind in zip(files, ids, kinds):
 
161
            f = f.strip('/')
 
162
            assert '//' not in f
 
163
            assert '..' not in f
 
164
            if self.path2id(f):
 
165
                # special case tree root handling.
 
166
                if f == '' and self.path2id(f) == ROOT_ID:
 
167
                    state.set_path_id('', generate_ids.gen_file_id(f))
 
168
                continue
 
169
            if file_id is None:
 
170
                file_id = generate_ids.gen_file_id(f)
 
171
            # deliberately add the file with no cached stat or sha1
 
172
            # - on the first access it will be gathered, and we can
 
173
            # always change this once tests are all passing.
 
174
            state.add(f, file_id, kind, None, '')
 
175
        self._make_dirty(reset_inventory=True)
 
176
 
 
177
    def _make_dirty(self, reset_inventory):
 
178
        """Make the tree state dirty.
 
179
 
 
180
        :param reset_inventory: True if the cached inventory should be removed
 
181
            (presuming there is one).
 
182
        """
 
183
        self._dirty = True
 
184
        if reset_inventory and self._inventory is not None:
 
185
            self._inventory = None
 
186
 
 
187
    @needs_tree_write_lock
 
188
    def add_reference(self, sub_tree):
 
189
        # use standard implementation, which calls back to self._add
 
190
        # 
 
191
        # So we don't store the reference_revision in the working dirstate,
 
192
        # it's just recorded at the moment of commit. 
 
193
        self._add_reference(sub_tree)
 
194
 
 
195
    def break_lock(self):
 
196
        """Break a lock if one is present from another instance.
 
197
 
 
198
        Uses the ui factory to ask for confirmation if the lock may be from
 
199
        an active process.
 
200
 
 
201
        This will probe the repository for its lock as well.
 
202
        """
 
203
        # if the dirstate is locked by an active process, reject the break lock
 
204
        # call.
 
205
        try:
 
206
            if self._dirstate is None:
 
207
                clear = True
 
208
            else:
 
209
                clear = False
 
210
            state = self._current_dirstate()
 
211
            if state._lock_token is not None:
 
212
                # we already have it locked. sheese, cant break our own lock.
 
213
                raise errors.LockActive(self.basedir)
 
214
            else:
 
215
                try:
 
216
                    # try for a write lock - need permission to get one anyhow
 
217
                    # to break locks.
 
218
                    state.lock_write()
 
219
                except errors.LockContention:
 
220
                    # oslocks fail when a process is still live: fail.
 
221
                    # TODO: get the locked lockdir info and give to the user to
 
222
                    # assist in debugging.
 
223
                    raise errors.LockActive(self.basedir)
 
224
                else:
 
225
                    state.unlock()
 
226
        finally:
 
227
            if clear:
 
228
                self._dirstate = None
 
229
        self._control_files.break_lock()
 
230
        self.branch.break_lock()
 
231
 
 
232
    def _comparison_data(self, entry, path):
 
233
        kind, executable, stat_value = \
 
234
            WorkingTree3._comparison_data(self, entry, path)
 
235
        # it looks like a plain directory, but it's really a reference -- see
 
236
        # also kind()
 
237
        if (self._repo_supports_tree_reference and
 
238
            kind == 'directory' and
 
239
            self._directory_is_tree_reference(path)):
 
240
            kind = 'tree-reference'
 
241
        return kind, executable, stat_value
 
242
 
 
243
    @needs_write_lock
 
244
    def commit(self, message=None, revprops=None, *args, **kwargs):
 
245
        # mark the tree as dirty post commit - commit
 
246
        # can change the current versioned list by doing deletes.
 
247
        result = WorkingTree3.commit(self, message, revprops, *args, **kwargs)
 
248
        self._make_dirty(reset_inventory=True)
 
249
        return result
 
250
 
 
251
    def current_dirstate(self):
 
252
        """Return the current dirstate object.
 
253
 
 
254
        This is not part of the tree interface and only exposed for ease of
 
255
        testing.
 
256
 
 
257
        :raises errors.NotWriteLocked: when not in a lock.
 
258
        """
 
259
        self._must_be_locked()
 
260
        return self._current_dirstate()
 
261
 
 
262
    def _current_dirstate(self):
 
263
        """Internal function that does not check lock status.
 
264
 
 
265
        This is needed for break_lock which also needs the dirstate.
 
266
        """
 
267
        if self._dirstate is not None:
 
268
            return self._dirstate
 
269
        local_path = self.bzrdir.get_workingtree_transport(None
 
270
            ).local_abspath('dirstate')
 
271
        self._dirstate = dirstate.DirState.on_file(local_path)
 
272
        return self._dirstate
 
273
 
 
274
    def _directory_is_tree_reference(self, relpath):
 
275
        # as a special case, if a directory contains control files then 
 
276
        # it's a tree reference, except that the root of the tree is not
 
277
        return relpath and osutils.isdir(self.abspath(relpath) + u"/.bzr")
 
278
        # TODO: We could ask all the control formats whether they
 
279
        # recognize this directory, but at the moment there's no cheap api
 
280
        # to do that.  Since we probably can only nest bzr checkouts and
 
281
        # they always use this name it's ok for now.  -- mbp 20060306
 
282
        #
 
283
        # FIXME: There is an unhandled case here of a subdirectory
 
284
        # containing .bzr but not a branch; that will probably blow up
 
285
        # when you try to commit it.  It might happen if there is a
 
286
        # checkout in a subdirectory.  This can be avoided by not adding
 
287
        # it.  mbp 20070306
 
288
 
 
289
    def filter_unversioned_files(self, paths):
 
290
        """Filter out paths that are versioned.
 
291
 
 
292
        :return: set of paths.
 
293
        """
 
294
        # TODO: make a generic multi-bisect routine roughly that should list
 
295
        # the paths, then process one half at a time recursively, and feed the
 
296
        # results of each bisect in further still
 
297
        paths = sorted(paths)
 
298
        result = set()
 
299
        state = self.current_dirstate()
 
300
        # TODO we want a paths_to_dirblocks helper I think
 
301
        for path in paths:
 
302
            dirname, basename = os.path.split(path.encode('utf8'))
 
303
            _, _, _, path_is_versioned = state._get_block_entry_index(
 
304
                dirname, basename, 0)
 
305
            if not path_is_versioned:
 
306
                result.add(path)
 
307
        return result
 
308
 
 
309
    def flush(self):
 
310
        """Write all cached data to disk."""
 
311
        if self._control_files._lock_mode != 'w':
 
312
            raise errors.NotWriteLocked(self)
 
313
        self.current_dirstate().save()
 
314
        self._inventory = None
 
315
        self._dirty = False
 
316
 
 
317
    @needs_tree_write_lock
 
318
    def _gather_kinds(self, files, kinds):
 
319
        """See MutableTree._gather_kinds."""
 
320
        for pos, f in enumerate(files):
 
321
            if kinds[pos] is None:
 
322
                kinds[pos] = self._kind(f)
 
323
 
 
324
    def _generate_inventory(self):
 
325
        """Create and set self.inventory from the dirstate object.
 
326
        
 
327
        This is relatively expensive: we have to walk the entire dirstate.
 
328
        Ideally we would not, and can deprecate this function.
 
329
        """
 
330
        #: uncomment to trap on inventory requests.
 
331
        # import pdb;pdb.set_trace()
 
332
        state = self.current_dirstate()
 
333
        state._read_dirblocks_if_needed()
 
334
        root_key, current_entry = self._get_entry(path='')
 
335
        current_id = root_key[2]
 
336
        assert current_entry[0][0] == 'd' # directory
 
337
        inv = Inventory(root_id=current_id)
 
338
        # Turn some things into local variables
 
339
        minikind_to_kind = dirstate.DirState._minikind_to_kind
 
340
        factory = entry_factory
 
341
        utf8_decode = cache_utf8._utf8_decode
 
342
        inv_byid = inv._byid
 
343
        # we could do this straight out of the dirstate; it might be fast
 
344
        # and should be profiled - RBC 20070216
 
345
        parent_ies = {'' : inv.root}
 
346
        for block in state._dirblocks[1:]: # skip the root
 
347
            dirname = block[0]
 
348
            try:
 
349
                parent_ie = parent_ies[dirname]
 
350
            except KeyError:
 
351
                # all the paths in this block are not versioned in this tree
 
352
                continue
 
353
            for key, entry in block[1]:
 
354
                minikind, link_or_sha1, size, executable, stat = entry[0]
 
355
                if minikind in ('a', 'r'): # absent, relocated
 
356
                    # a parent tree only entry
 
357
                    continue
 
358
                name = key[1]
 
359
                name_unicode = utf8_decode(name)[0]
 
360
                file_id = key[2]
 
361
                kind = minikind_to_kind[minikind]
 
362
                inv_entry = factory[kind](file_id, name_unicode,
 
363
                                          parent_ie.file_id)
 
364
                if kind == 'file':
 
365
                    # This is only needed on win32, where this is the only way
 
366
                    # we know the executable bit.
 
367
                    inv_entry.executable = executable
 
368
                    # not strictly needed: working tree
 
369
                    #inv_entry.text_size = size
 
370
                    #inv_entry.text_sha1 = sha1
 
371
                elif kind == 'directory':
 
372
                    # add this entry to the parent map.
 
373
                    parent_ies[(dirname + '/' + name).strip('/')] = inv_entry
 
374
                elif kind == 'tree-reference':
 
375
                    assert self._repo_supports_tree_reference, \
 
376
                        "repository of %r " \
 
377
                        "doesn't support tree references " \
 
378
                        "required by entry %r" \
 
379
                        % (self, name)
 
380
                    inv_entry.reference_revision = link_or_sha1 or None
 
381
                elif kind != 'symlink':
 
382
                    raise AssertionError("unknown kind %r" % kind)
 
383
                # These checks cost us around 40ms on a 55k entry tree
 
384
                assert file_id not in inv_byid, ('file_id %s already in'
 
385
                    ' inventory as %s' % (file_id, inv_byid[file_id]))
 
386
                assert name_unicode not in parent_ie.children
 
387
                inv_byid[file_id] = inv_entry
 
388
                parent_ie.children[name_unicode] = inv_entry
 
389
        self._inventory = inv
 
390
 
 
391
    def _get_entry(self, file_id=None, path=None):
 
392
        """Get the dirstate row for file_id or path.
 
393
 
 
394
        If either file_id or path is supplied, it is used as the key to lookup.
 
395
        If both are supplied, the fastest lookup is used, and an error is
 
396
        raised if they do not both point at the same row.
 
397
        
 
398
        :param file_id: An optional unicode file_id to be looked up.
 
399
        :param path: An optional unicode path to be looked up.
 
400
        :return: The dirstate row tuple for path/file_id, or (None, None)
 
401
        """
 
402
        if file_id is None and path is None:
 
403
            raise errors.BzrError('must supply file_id or path')
 
404
        state = self.current_dirstate()
 
405
        if path is not None:
 
406
            path = path.encode('utf8')
 
407
        return state._get_entry(0, fileid_utf8=file_id, path_utf8=path)
 
408
 
 
409
    def get_file_sha1(self, file_id, path=None, stat_value=None):
 
410
        # check file id is valid unconditionally.
 
411
        entry = self._get_entry(file_id=file_id, path=path)
 
412
        if entry[0] is None:
 
413
            raise errors.NoSuchId(self, file_id)
 
414
        if path is None:
 
415
            path = pathjoin(entry[0][0], entry[0][1]).decode('utf8')
 
416
 
 
417
        file_abspath = self.abspath(path)
 
418
        state = self.current_dirstate()
 
419
        if stat_value is None:
 
420
            try:
 
421
                stat_value = os.lstat(file_abspath)
 
422
            except OSError, e:
 
423
                if e.errno == errno.ENOENT:
 
424
                    return None
 
425
                else:
 
426
                    raise
 
427
        link_or_sha1 = state.update_entry(entry, file_abspath,
 
428
                                          stat_value=stat_value)
 
429
        if entry[1][0][0] == 'f':
 
430
            return link_or_sha1
 
431
        return None
 
432
 
 
433
    def _get_inventory(self):
 
434
        """Get the inventory for the tree. This is only valid within a lock."""
 
435
        if 'evil' in debug.debug_flags:
 
436
            trace.mutter_callsite(2,
 
437
                "accessing .inventory forces a size of tree translation.")
 
438
        if self._inventory is not None:
 
439
            return self._inventory
 
440
        self._must_be_locked()
 
441
        self._generate_inventory()
 
442
        return self._inventory
 
443
 
 
444
    inventory = property(_get_inventory,
 
445
                         doc="Inventory of this Tree")
 
446
 
 
447
    @needs_read_lock
 
448
    def get_parent_ids(self):
 
449
        """See Tree.get_parent_ids.
 
450
        
 
451
        This implementation requests the ids list from the dirstate file.
 
452
        """
 
453
        return self.current_dirstate().get_parent_ids()
 
454
 
 
455
    def get_reference_revision(self, file_id, path=None):
 
456
        # referenced tree's revision is whatever's currently there
 
457
        return self.get_nested_tree(file_id, path).last_revision()
 
458
 
 
459
    def get_nested_tree(self, file_id, path=None):
 
460
        if path is None:
 
461
            path = self.id2path(file_id)
 
462
        # else: check file_id is at path?
 
463
        return WorkingTree.open(self.abspath(path))
 
464
 
 
465
    @needs_read_lock
 
466
    def get_root_id(self):
 
467
        """Return the id of this trees root"""
 
468
        return self._get_entry(path='')[0][2]
 
469
 
 
470
    def has_id(self, file_id):
 
471
        state = self.current_dirstate()
 
472
        file_id = osutils.safe_file_id(file_id)
 
473
        row, parents = self._get_entry(file_id=file_id)
 
474
        if row is None:
 
475
            return False
 
476
        return osutils.lexists(pathjoin(
 
477
                    self.basedir, row[0].decode('utf8'), row[1].decode('utf8')))
 
478
 
 
479
    @needs_read_lock
 
480
    def id2path(self, file_id):
 
481
        "Convert a file-id to a path."
 
482
        file_id = osutils.safe_file_id(file_id)
 
483
        state = self.current_dirstate()
 
484
        entry = self._get_entry(file_id=file_id)
 
485
        if entry == (None, None):
 
486
            raise errors.NoSuchId(tree=self, file_id=file_id)
 
487
        path_utf8 = osutils.pathjoin(entry[0][0], entry[0][1])
 
488
        return path_utf8.decode('utf8')
 
489
 
 
490
    if not osutils.supports_executable():
 
491
        def is_executable(self, file_id, path=None):
 
492
            """Test if a file is executable or not.
 
493
 
 
494
            Note: The caller is expected to take a read-lock before calling this.
 
495
            """
 
496
            file_id = osutils.safe_file_id(file_id)
 
497
            entry = self._get_entry(file_id=file_id, path=path)
 
498
            if entry == (None, None):
 
499
                return False
 
500
            return entry[1][0][3]
 
501
    else:
 
502
        def is_executable(self, file_id, path=None):
 
503
            """Test if a file is executable or not.
 
504
 
 
505
            Note: The caller is expected to take a read-lock before calling this.
 
506
            """
 
507
            if not path:
 
508
                file_id = osutils.safe_file_id(file_id)
 
509
                path = self.id2path(file_id)
 
510
            mode = os.lstat(self.abspath(path)).st_mode
 
511
            return bool(stat.S_ISREG(mode) and stat.S_IEXEC & mode)
 
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
                        extra="(Use --after to update the Bazaar id)")
 
738
 
 
739
            rollbacks = []
 
740
            def rollback_rename():
 
741
                """A single rename has failed, roll it back."""
 
742
                # roll back everything, even if we encounter trouble doing one
 
743
                # of them.
 
744
                #
 
745
                # TODO: at least log the other exceptions rather than just
 
746
                # losing them mbp 20070307
 
747
                exc_info = None
 
748
                for rollback in reversed(rollbacks):
 
749
                    try:
 
750
                        rollback()
 
751
                    except Exception, e:
 
752
                        exc_info = sys.exc_info()
 
753
                if exc_info:
 
754
                    raise exc_info[0], exc_info[1], exc_info[2]
 
755
 
 
756
            # perform the disk move first - its the most likely failure point.
 
757
            if move_file:
 
758
                from_rel_abs = self.abspath(from_rel)
 
759
                to_rel_abs = self.abspath(to_rel)
 
760
                try:
 
761
                    osutils.rename(from_rel_abs, to_rel_abs)
 
762
                except OSError, e:
 
763
                    raise errors.BzrMoveFailedError(from_rel, to_rel, e[1])
 
764
                rollbacks.append(lambda: osutils.rename(to_rel_abs, from_rel_abs))
 
765
            try:
 
766
                # perform the rename in the inventory next if needed: its easy
 
767
                # to rollback
 
768
                if update_inventory:
 
769
                    # rename the entry
 
770
                    from_entry = inv[from_id]
 
771
                    current_parent = from_entry.parent_id
 
772
                    inv.rename(from_id, to_dir_id, from_tail)
 
773
                    rollbacks.append(
 
774
                        lambda: inv.rename(from_id, current_parent, from_tail))
 
775
                # finally do the rename in the dirstate, which is a little
 
776
                # tricky to rollback, but least likely to need it.
 
777
                old_block_index, old_entry_index, dir_present, file_present = \
 
778
                    state._get_block_entry_index(from_dirname, from_tail_utf8, 0)
 
779
                old_block = state._dirblocks[old_block_index][1]
 
780
                old_entry = old_block[old_entry_index]
 
781
                from_key, old_entry_details = old_entry
 
782
                cur_details = old_entry_details[0]
 
783
                # remove the old row
 
784
                to_key = ((to_block[0],) + from_key[1:3])
 
785
                minikind = cur_details[0]
 
786
                move_one(old_entry, from_path_utf8=from_rel_utf8,
 
787
                         minikind=minikind,
 
788
                         executable=cur_details[3],
 
789
                         fingerprint=cur_details[1],
 
790
                         packed_stat=cur_details[4],
 
791
                         size=cur_details[2],
 
792
                         to_block=to_block,
 
793
                         to_key=to_key,
 
794
                         to_path_utf8=to_rel_utf8)
 
795
 
 
796
                if minikind == 'd':
 
797
                    def update_dirblock(from_dir, to_key, to_dir_utf8):
 
798
                        """Recursively update all entries in this dirblock."""
 
799
                        assert from_dir != '', "renaming root not supported"
 
800
                        from_key = (from_dir, '')
 
801
                        from_block_idx, present = \
 
802
                            state._find_block_index_from_key(from_key)
 
803
                        if not present:
 
804
                            # This is the old record, if it isn't present, then
 
805
                            # there is theoretically nothing to update.
 
806
                            # (Unless it isn't present because of lazy loading,
 
807
                            # but we don't do that yet)
 
808
                            return
 
809
                        from_block = state._dirblocks[from_block_idx]
 
810
                        to_block_index, to_entry_index, _, _ = \
 
811
                            state._get_block_entry_index(to_key[0], to_key[1], 0)
 
812
                        to_block_index = state._ensure_block(
 
813
                            to_block_index, to_entry_index, to_dir_utf8)
 
814
                        to_block = state._dirblocks[to_block_index]
 
815
 
 
816
                        # Grab a copy since move_one may update the list.
 
817
                        for entry in from_block[1][:]:
 
818
                            assert entry[0][0] == from_dir
 
819
                            cur_details = entry[1][0]
 
820
                            to_key = (to_dir_utf8, entry[0][1], entry[0][2])
 
821
                            from_path_utf8 = osutils.pathjoin(entry[0][0], entry[0][1])
 
822
                            to_path_utf8 = osutils.pathjoin(to_dir_utf8, entry[0][1])
 
823
                            minikind = cur_details[0]
 
824
                            if minikind in 'ar':
 
825
                                # Deleted children of a renamed directory
 
826
                                # Do not need to be updated.
 
827
                                # Children that have been renamed out of this
 
828
                                # directory should also not be updated
 
829
                                continue
 
830
                            move_one(entry, from_path_utf8=from_path_utf8,
 
831
                                     minikind=minikind,
 
832
                                     executable=cur_details[3],
 
833
                                     fingerprint=cur_details[1],
 
834
                                     packed_stat=cur_details[4],
 
835
                                     size=cur_details[2],
 
836
                                     to_block=to_block,
 
837
                                     to_key=to_key,
 
838
                                     to_path_utf8=to_path_utf8)
 
839
                            if minikind == 'd':
 
840
                                # We need to move all the children of this
 
841
                                # entry
 
842
                                update_dirblock(from_path_utf8, to_key,
 
843
                                                to_path_utf8)
 
844
                    update_dirblock(from_rel_utf8, to_key, to_rel_utf8)
 
845
            except:
 
846
                rollback_rename()
 
847
                raise
 
848
            result.append((from_rel, to_rel))
 
849
            state._dirblock_state = dirstate.DirState.IN_MEMORY_MODIFIED
 
850
            self._make_dirty(reset_inventory=False)
 
851
 
 
852
        return result
 
853
 
 
854
    def _must_be_locked(self):
 
855
        if not self._control_files._lock_count:
 
856
            raise errors.ObjectNotLocked(self)
 
857
 
 
858
    def _new_tree(self):
 
859
        """Initialize the state in this tree to be a new tree."""
 
860
        self._dirty = True
 
861
 
 
862
    @needs_read_lock
 
863
    def path2id(self, path):
 
864
        """Return the id for path in this tree."""
 
865
        path = path.strip('/')
 
866
        entry = self._get_entry(path=path)
 
867
        if entry == (None, None):
 
868
            return None
 
869
        return entry[0][2]
 
870
 
 
871
    def paths2ids(self, paths, trees=[], require_versioned=True):
 
872
        """See Tree.paths2ids().
 
873
 
 
874
        This specialisation fast-paths the case where all the trees are in the
 
875
        dirstate.
 
876
        """
 
877
        if paths is None:
 
878
            return None
 
879
        parents = self.get_parent_ids()
 
880
        for tree in trees:
 
881
            if not (isinstance(tree, DirStateRevisionTree) and tree._revision_id in
 
882
                parents):
 
883
                return super(WorkingTree4, self).paths2ids(paths, trees, require_versioned)
 
884
        search_indexes = [0] + [1 + parents.index(tree._revision_id) for tree in trees]
 
885
        # -- make all paths utf8 --
 
886
        paths_utf8 = set()
 
887
        for path in paths:
 
888
            paths_utf8.add(path.encode('utf8'))
 
889
        paths = paths_utf8
 
890
        # -- paths is now a utf8 path set --
 
891
        # -- get the state object and prepare it.
 
892
        state = self.current_dirstate()
 
893
        if False and (state._dirblock_state == dirstate.DirState.NOT_IN_MEMORY
 
894
            and '' not in paths):
 
895
            paths2ids = self._paths2ids_using_bisect
 
896
        else:
 
897
            paths2ids = self._paths2ids_in_memory
 
898
        return paths2ids(paths, search_indexes,
 
899
                         require_versioned=require_versioned)
 
900
 
 
901
    def _paths2ids_in_memory(self, paths, search_indexes,
 
902
                             require_versioned=True):
 
903
        state = self.current_dirstate()
 
904
        state._read_dirblocks_if_needed()
 
905
        def _entries_for_path(path):
 
906
            """Return a list with all the entries that match path for all ids.
 
907
            """
 
908
            dirname, basename = os.path.split(path)
 
909
            key = (dirname, basename, '')
 
910
            block_index, present = state._find_block_index_from_key(key)
 
911
            if not present:
 
912
                # the block which should contain path is absent.
 
913
                return []
 
914
            result = []
 
915
            block = state._dirblocks[block_index][1]
 
916
            entry_index, _ = state._find_entry_index(key, block)
 
917
            # we may need to look at multiple entries at this path: walk while the paths match.
 
918
            while (entry_index < len(block) and
 
919
                block[entry_index][0][0:2] == key[0:2]):
 
920
                result.append(block[entry_index])
 
921
                entry_index += 1
 
922
            return result
 
923
        if require_versioned:
 
924
            # -- check all supplied paths are versioned in a search tree. --
 
925
            all_versioned = True
 
926
            for path in paths:
 
927
                path_entries = _entries_for_path(path)
 
928
                if not path_entries:
 
929
                    # this specified path is not present at all: error
 
930
                    all_versioned = False
 
931
                    break
 
932
                found_versioned = False
 
933
                # for each id at this path
 
934
                for entry in path_entries:
 
935
                    # for each tree.
 
936
                    for index in search_indexes:
 
937
                        if entry[1][index][0] != 'a': # absent
 
938
                            found_versioned = True
 
939
                            # all good: found a versioned cell
 
940
                            break
 
941
                if not found_versioned:
 
942
                    # none of the indexes was not 'absent' at all ids for this
 
943
                    # path.
 
944
                    all_versioned = False
 
945
                    break
 
946
            if not all_versioned:
 
947
                raise errors.PathsNotVersionedError(paths)
 
948
        # -- remove redundancy in supplied paths to prevent over-scanning --
 
949
        search_paths = osutils.minimum_path_selection(paths)
 
950
        # sketch: 
 
951
        # for all search_indexs in each path at or under each element of
 
952
        # search_paths, if the detail is relocated: add the id, and add the
 
953
        # relocated path as one to search if its not searched already. If the
 
954
        # detail is not relocated, add the id.
 
955
        searched_paths = set()
 
956
        found_ids = set()
 
957
        def _process_entry(entry):
 
958
            """Look at search_indexes within entry.
 
959
 
 
960
            If a specific tree's details are relocated, add the relocation
 
961
            target to search_paths if not searched already. If it is absent, do
 
962
            nothing. Otherwise add the id to found_ids.
 
963
            """
 
964
            for index in search_indexes:
 
965
                if entry[1][index][0] == 'r': # relocated
 
966
                    if not osutils.is_inside_any(searched_paths, entry[1][index][1]):
 
967
                        search_paths.add(entry[1][index][1])
 
968
                elif entry[1][index][0] != 'a': # absent
 
969
                    found_ids.add(entry[0][2])
 
970
        while search_paths:
 
971
            current_root = search_paths.pop()
 
972
            searched_paths.add(current_root)
 
973
            # process the entries for this containing directory: the rest will be
 
974
            # found by their parents recursively.
 
975
            root_entries = _entries_for_path(current_root)
 
976
            if not root_entries:
 
977
                # this specified path is not present at all, skip it.
 
978
                continue
 
979
            for entry in root_entries:
 
980
                _process_entry(entry)
 
981
            initial_key = (current_root, '', '')
 
982
            block_index, _ = state._find_block_index_from_key(initial_key)
 
983
            while (block_index < len(state._dirblocks) and
 
984
                osutils.is_inside(current_root, state._dirblocks[block_index][0])):
 
985
                for entry in state._dirblocks[block_index][1]:
 
986
                    _process_entry(entry)
 
987
                block_index += 1
 
988
        return found_ids
 
989
 
 
990
    def _paths2ids_using_bisect(self, paths, search_indexes,
 
991
                                require_versioned=True):
 
992
        state = self.current_dirstate()
 
993
        found_ids = set()
 
994
 
 
995
        split_paths = sorted(osutils.split(p) for p in paths)
 
996
        found = state._bisect_recursive(split_paths)
 
997
 
 
998
        if require_versioned:
 
999
            found_dir_names = set(dir_name_id[:2] for dir_name_id in found)
 
1000
            for dir_name in split_paths:
 
1001
                if dir_name not in found_dir_names:
 
1002
                    raise errors.PathsNotVersionedError(paths)
 
1003
 
 
1004
        for dir_name_id, trees_info in found.iteritems():
 
1005
            for index in search_indexes:
 
1006
                if trees_info[index][0] not in ('r', 'a'):
 
1007
                    found_ids.add(dir_name_id[2])
 
1008
        return found_ids
 
1009
 
 
1010
    def read_working_inventory(self):
 
1011
        """Read the working inventory.
 
1012
        
 
1013
        This is a meaningless operation for dirstate, but we obey it anyhow.
 
1014
        """
 
1015
        return self.inventory
 
1016
 
 
1017
    @needs_read_lock
 
1018
    def revision_tree(self, revision_id):
 
1019
        """See Tree.revision_tree.
 
1020
 
 
1021
        WorkingTree4 supplies revision_trees for any basis tree.
 
1022
        """
 
1023
        revision_id = osutils.safe_revision_id(revision_id)
 
1024
        dirstate = self.current_dirstate()
 
1025
        parent_ids = dirstate.get_parent_ids()
 
1026
        if revision_id not in parent_ids:
 
1027
            raise errors.NoSuchRevisionInTree(self, revision_id)
 
1028
        if revision_id in dirstate.get_ghosts():
 
1029
            raise errors.NoSuchRevisionInTree(self, revision_id)
 
1030
        return DirStateRevisionTree(dirstate, revision_id,
 
1031
            self.branch.repository)
 
1032
 
 
1033
    @needs_tree_write_lock
 
1034
    def set_last_revision(self, new_revision):
 
1035
        """Change the last revision in the working tree."""
 
1036
        new_revision = osutils.safe_revision_id(new_revision)
 
1037
        parents = self.get_parent_ids()
 
1038
        if new_revision in (NULL_REVISION, None):
 
1039
            assert len(parents) < 2, (
 
1040
                "setting the last parent to none with a pending merge is "
 
1041
                "unsupported.")
 
1042
            self.set_parent_ids([])
 
1043
        else:
 
1044
            self.set_parent_ids([new_revision] + parents[1:],
 
1045
                allow_leftmost_as_ghost=True)
 
1046
 
 
1047
    @needs_tree_write_lock
 
1048
    def set_parent_ids(self, revision_ids, allow_leftmost_as_ghost=False):
 
1049
        """Set the parent ids to revision_ids.
 
1050
        
 
1051
        See also set_parent_trees. This api will try to retrieve the tree data
 
1052
        for each element of revision_ids from the trees repository. If you have
 
1053
        tree data already available, it is more efficient to use
 
1054
        set_parent_trees rather than set_parent_ids. set_parent_ids is however
 
1055
        an easier API to use.
 
1056
 
 
1057
        :param revision_ids: The revision_ids to set as the parent ids of this
 
1058
            working tree. Any of these may be ghosts.
 
1059
        """
 
1060
        revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
 
1061
        trees = []
 
1062
        for revision_id in revision_ids:
 
1063
            try:
 
1064
                revtree = self.branch.repository.revision_tree(revision_id)
 
1065
                # TODO: jam 20070213 KnitVersionedFile raises
 
1066
                #       RevisionNotPresent rather than NoSuchRevision if a
 
1067
                #       given revision_id is not present. Should Repository be
 
1068
                #       catching it and re-raising NoSuchRevision?
 
1069
            except (errors.NoSuchRevision, errors.RevisionNotPresent):
 
1070
                revtree = None
 
1071
            trees.append((revision_id, revtree))
 
1072
        self.set_parent_trees(trees,
 
1073
            allow_leftmost_as_ghost=allow_leftmost_as_ghost)
 
1074
 
 
1075
    @needs_tree_write_lock
 
1076
    def set_parent_trees(self, parents_list, allow_leftmost_as_ghost=False):
 
1077
        """Set the parents of the working tree.
 
1078
 
 
1079
        :param parents_list: A list of (revision_id, tree) tuples.
 
1080
            If tree is None, then that element is treated as an unreachable
 
1081
            parent tree - i.e. a ghost.
 
1082
        """
 
1083
        dirstate = self.current_dirstate()
 
1084
        if len(parents_list) > 0:
 
1085
            if not allow_leftmost_as_ghost and parents_list[0][1] is None:
 
1086
                raise errors.GhostRevisionUnusableHere(parents_list[0][0])
 
1087
        real_trees = []
 
1088
        ghosts = []
 
1089
        # convert absent trees to the null tree, which we convert back to
 
1090
        # missing on access.
 
1091
        for rev_id, tree in parents_list:
 
1092
            rev_id = osutils.safe_revision_id(rev_id)
 
1093
            _mod_revision.check_not_reserved_id(rev_id)
 
1094
            if tree is not None:
 
1095
                real_trees.append((rev_id, tree))
 
1096
            else:
 
1097
                real_trees.append((rev_id,
 
1098
                    self.branch.repository.revision_tree(None)))
 
1099
                ghosts.append(rev_id)
 
1100
        dirstate.set_parent_trees(real_trees, ghosts=ghosts)
 
1101
        self._make_dirty(reset_inventory=False)
 
1102
 
 
1103
    def _set_root_id(self, file_id):
 
1104
        """See WorkingTree.set_root_id."""
 
1105
        state = self.current_dirstate()
 
1106
        state.set_path_id('', file_id)
 
1107
        if state._dirblock_state == dirstate.DirState.IN_MEMORY_MODIFIED:
 
1108
            self._make_dirty(reset_inventory=True)
 
1109
 
 
1110
    @needs_read_lock
 
1111
    def supports_tree_reference(self):
 
1112
        return self._repo_supports_tree_reference
 
1113
 
 
1114
    def unlock(self):
 
1115
        """Unlock in format 4 trees needs to write the entire dirstate."""
 
1116
        # do non-implementation specific cleanup
 
1117
        self._cleanup()
 
1118
 
 
1119
        if self._control_files._lock_count == 1:
 
1120
            # eventually we should do signature checking during read locks for
 
1121
            # dirstate updates.
 
1122
            if self._control_files._lock_mode == 'w':
 
1123
                if self._dirty:
 
1124
                    self.flush()
 
1125
            if self._dirstate is not None:
 
1126
                # This is a no-op if there are no modifications.
 
1127
                self._dirstate.save()
 
1128
                self._dirstate.unlock()
 
1129
            # TODO: jam 20070301 We shouldn't have to wipe the dirstate at this
 
1130
            #       point. Instead, it could check if the header has been
 
1131
            #       modified when it is locked, and if not, it can hang on to
 
1132
            #       the data it has in memory.
 
1133
            self._dirstate = None
 
1134
            self._inventory = None
 
1135
        # reverse order of locking.
 
1136
        try:
 
1137
            return self._control_files.unlock()
 
1138
        finally:
 
1139
            self.branch.unlock()
 
1140
 
 
1141
    @needs_tree_write_lock
 
1142
    def unversion(self, file_ids):
 
1143
        """Remove the file ids in file_ids from the current versioned set.
 
1144
 
 
1145
        When a file_id is unversioned, all of its children are automatically
 
1146
        unversioned.
 
1147
 
 
1148
        :param file_ids: The file ids to stop versioning.
 
1149
        :raises: NoSuchId if any fileid is not currently versioned.
 
1150
        """
 
1151
        if not file_ids:
 
1152
            return
 
1153
        state = self.current_dirstate()
 
1154
        state._read_dirblocks_if_needed()
 
1155
        ids_to_unversion = set()
 
1156
        for file_id in file_ids:
 
1157
            ids_to_unversion.add(osutils.safe_file_id(file_id))
 
1158
        paths_to_unversion = set()
 
1159
        # sketch:
 
1160
        # check if the root is to be unversioned, if so, assert for now.
 
1161
        # walk the state marking unversioned things as absent.
 
1162
        # if there are any un-unversioned ids at the end, raise
 
1163
        for key, details in state._dirblocks[0][1]:
 
1164
            if (details[0][0] not in ('a', 'r') and # absent or relocated
 
1165
                key[2] in ids_to_unversion):
 
1166
                # I haven't written the code to unversion / yet - it should be
 
1167
                # supported.
 
1168
                raise errors.BzrError('Unversioning the / is not currently supported')
 
1169
        block_index = 0
 
1170
        while block_index < len(state._dirblocks):
 
1171
            # process one directory at a time.
 
1172
            block = state._dirblocks[block_index]
 
1173
            # first check: is the path one to remove - it or its children
 
1174
            delete_block = False
 
1175
            for path in paths_to_unversion:
 
1176
                if (block[0].startswith(path) and
 
1177
                    (len(block[0]) == len(path) or
 
1178
                     block[0][len(path)] == '/')):
 
1179
                    # this entire block should be deleted - its the block for a
 
1180
                    # path to unversion; or the child of one
 
1181
                    delete_block = True
 
1182
                    break
 
1183
            # TODO: trim paths_to_unversion as we pass by paths
 
1184
            if delete_block:
 
1185
                # this block is to be deleted: process it.
 
1186
                # TODO: we can special case the no-parents case and
 
1187
                # just forget the whole block.
 
1188
                entry_index = 0
 
1189
                while entry_index < len(block[1]):
 
1190
                    # Mark this file id as having been removed
 
1191
                    entry = block[1][entry_index]
 
1192
                    ids_to_unversion.discard(entry[0][2])
 
1193
                    if (entry[1][0][0] == 'a'
 
1194
                        or not state._make_absent(entry)):
 
1195
                        entry_index += 1
 
1196
                # go to the next block. (At the moment we dont delete empty
 
1197
                # dirblocks)
 
1198
                block_index += 1
 
1199
                continue
 
1200
            entry_index = 0
 
1201
            while entry_index < len(block[1]):
 
1202
                entry = block[1][entry_index]
 
1203
                if (entry[1][0][0] in ('a', 'r') or # absent, relocated
 
1204
                    # ^ some parent row.
 
1205
                    entry[0][2] not in ids_to_unversion):
 
1206
                    # ^ not an id to unversion
 
1207
                    entry_index += 1
 
1208
                    continue
 
1209
                if entry[1][0][0] == 'd':
 
1210
                    paths_to_unversion.add(pathjoin(entry[0][0], entry[0][1]))
 
1211
                if not state._make_absent(entry):
 
1212
                    entry_index += 1
 
1213
                # we have unversioned this id
 
1214
                ids_to_unversion.remove(entry[0][2])
 
1215
            block_index += 1
 
1216
        if ids_to_unversion:
 
1217
            raise errors.NoSuchId(self, iter(ids_to_unversion).next())
 
1218
        self._make_dirty(reset_inventory=False)
 
1219
        # have to change the legacy inventory too.
 
1220
        if self._inventory is not None:
 
1221
            for file_id in file_ids:
 
1222
                self._inventory.remove_recursive_id(file_id)
 
1223
 
 
1224
    @needs_read_lock
 
1225
    def _validate(self):
 
1226
        self._dirstate._validate()
 
1227
 
 
1228
    @needs_tree_write_lock
 
1229
    def _write_inventory(self, inv):
 
1230
        """Write inventory as the current inventory."""
 
1231
        assert not self._dirty, "attempting to write an inventory when the dirstate is dirty will cause data loss"
 
1232
        self.current_dirstate().set_state_from_inventory(inv)
 
1233
        self._make_dirty(reset_inventory=False)
 
1234
        if self._inventory is not None:
 
1235
            self._inventory = inv
 
1236
        self.flush()
 
1237
 
 
1238
 
 
1239
class WorkingTreeFormat4(WorkingTreeFormat3):
 
1240
    """The first consolidated dirstate working tree format.
 
1241
 
 
1242
    This format:
 
1243
        - exists within a metadir controlling .bzr
 
1244
        - includes an explicit version marker for the workingtree control
 
1245
          files, separate from the BzrDir format
 
1246
        - modifies the hash cache format
 
1247
        - is new in bzr 0.15
 
1248
        - uses a LockDir to guard access to it.
 
1249
    """
 
1250
 
 
1251
    upgrade_recommended = False
 
1252
 
 
1253
    def get_format_string(self):
 
1254
        """See WorkingTreeFormat.get_format_string()."""
 
1255
        return "Bazaar Working Tree Format 4 (bzr 0.15)\n"
 
1256
 
 
1257
    def get_format_description(self):
 
1258
        """See WorkingTreeFormat.get_format_description()."""
 
1259
        return "Working tree format 4"
 
1260
 
 
1261
    def initialize(self, a_bzrdir, revision_id=None):
 
1262
        """See WorkingTreeFormat.initialize().
 
1263
 
 
1264
        :param revision_id: allows creating a working tree at a different
 
1265
        revision than the branch is at.
 
1266
 
 
1267
        These trees get an initial random root id, if their repository supports
 
1268
        rich root data, TREE_ROOT otherwise.
 
1269
        """
 
1270
        revision_id = osutils.safe_revision_id(revision_id)
 
1271
        if not isinstance(a_bzrdir.transport, LocalTransport):
 
1272
            raise errors.NotLocalUrl(a_bzrdir.transport.base)
 
1273
        transport = a_bzrdir.get_workingtree_transport(self)
 
1274
        control_files = self._open_control_files(a_bzrdir)
 
1275
        control_files.create_lock()
 
1276
        control_files.lock_write()
 
1277
        control_files.put_utf8('format', self.get_format_string())
 
1278
        branch = a_bzrdir.open_branch()
 
1279
        if revision_id is None:
 
1280
            revision_id = branch.last_revision()
 
1281
        local_path = transport.local_abspath('dirstate')
 
1282
        # write out new dirstate (must exist when we create the tree)
 
1283
        state = dirstate.DirState.initialize(local_path)
 
1284
        state.unlock()
 
1285
        del state
 
1286
        wt = WorkingTree4(a_bzrdir.root_transport.local_abspath('.'),
 
1287
                         branch,
 
1288
                         _format=self,
 
1289
                         _bzrdir=a_bzrdir,
 
1290
                         _control_files=control_files)
 
1291
        wt._new_tree()
 
1292
        wt.lock_tree_write()
 
1293
        try:
 
1294
            if revision_id in (None, NULL_REVISION):
 
1295
                if branch.repository.supports_rich_root():
 
1296
                    wt._set_root_id(generate_ids.gen_root_id())
 
1297
                else:
 
1298
                    wt._set_root_id(ROOT_ID)
 
1299
                wt.flush()
 
1300
            wt.set_last_revision(revision_id)
 
1301
            wt.flush()
 
1302
            basis = wt.basis_tree()
 
1303
            basis.lock_read()
 
1304
            # if the basis has a root id we have to use that; otherwise we use
 
1305
            # a new random one
 
1306
            basis_root_id = basis.get_root_id()
 
1307
            if basis_root_id is not None:
 
1308
                wt._set_root_id(basis_root_id)
 
1309
                wt.flush()
 
1310
            transform.build_tree(basis, wt)
 
1311
            basis.unlock()
 
1312
        finally:
 
1313
            control_files.unlock()
 
1314
            wt.unlock()
 
1315
        return wt
 
1316
 
 
1317
    def _open(self, a_bzrdir, control_files):
 
1318
        """Open the tree itself.
 
1319
 
 
1320
        :param a_bzrdir: the dir for the tree.
 
1321
        :param control_files: the control files for the tree.
 
1322
        """
 
1323
        return WorkingTree4(a_bzrdir.root_transport.local_abspath('.'),
 
1324
                           branch=a_bzrdir.open_branch(),
 
1325
                           _format=self,
 
1326
                           _bzrdir=a_bzrdir,
 
1327
                           _control_files=control_files)
 
1328
 
 
1329
    def __get_matchingbzrdir(self):
 
1330
        # please test against something that will let us do tree references
 
1331
        return bzrdir.format_registry.make_bzrdir(
 
1332
            'dirstate-with-subtree')
 
1333
 
 
1334
    _matchingbzrdir = property(__get_matchingbzrdir)
 
1335
 
 
1336
 
 
1337
class DirStateRevisionTree(Tree):
 
1338
    """A revision tree pulling the inventory from a dirstate."""
 
1339
 
 
1340
    def __init__(self, dirstate, revision_id, repository):
 
1341
        self._dirstate = dirstate
 
1342
        self._revision_id = osutils.safe_revision_id(revision_id)
 
1343
        self._repository = repository
 
1344
        self._inventory = None
 
1345
        self._locked = 0
 
1346
        self._dirstate_locked = False
 
1347
 
 
1348
    def __repr__(self):
 
1349
        return "<%s of %s in %s>" % \
 
1350
            (self.__class__.__name__, self._revision_id, self._dirstate)
 
1351
 
 
1352
    def annotate_iter(self, file_id,
 
1353
                      default_revision=_mod_revision.CURRENT_REVISION):
 
1354
        """See Tree.annotate_iter"""
 
1355
        w = self._get_weave(file_id)
 
1356
        return w.annotate_iter(self.inventory[file_id].revision)
 
1357
 
 
1358
    def _get_ancestors(self, default_revision):
 
1359
        return set(self._repository.get_ancestry(self._revision_id,
 
1360
                                                 topo_sorted=False))
 
1361
    def _comparison_data(self, entry, path):
 
1362
        """See Tree._comparison_data."""
 
1363
        if entry is None:
 
1364
            return None, False, None
 
1365
        # trust the entry as RevisionTree does, but this may not be
 
1366
        # sensible: the entry might not have come from us?
 
1367
        return entry.kind, entry.executable, None
 
1368
 
 
1369
    def _file_size(self, entry, stat_value):
 
1370
        return entry.text_size
 
1371
 
 
1372
    def filter_unversioned_files(self, paths):
 
1373
        """Filter out paths that are not versioned.
 
1374
 
 
1375
        :return: set of paths.
 
1376
        """
 
1377
        pred = self.has_filename
 
1378
        return set((p for p in paths if not pred(p)))
 
1379
 
 
1380
    def get_root_id(self):
 
1381
        return self.path2id('')
 
1382
 
 
1383
    def _get_parent_index(self):
 
1384
        """Return the index in the dirstate referenced by this tree."""
 
1385
        return self._dirstate.get_parent_ids().index(self._revision_id) + 1
 
1386
 
 
1387
    def _get_entry(self, file_id=None, path=None):
 
1388
        """Get the dirstate row for file_id or path.
 
1389
 
 
1390
        If either file_id or path is supplied, it is used as the key to lookup.
 
1391
        If both are supplied, the fastest lookup is used, and an error is
 
1392
        raised if they do not both point at the same row.
 
1393
        
 
1394
        :param file_id: An optional unicode file_id to be looked up.
 
1395
        :param path: An optional unicode path to be looked up.
 
1396
        :return: The dirstate row tuple for path/file_id, or (None, None)
 
1397
        """
 
1398
        if file_id is None and path is None:
 
1399
            raise errors.BzrError('must supply file_id or path')
 
1400
        file_id = osutils.safe_file_id(file_id)
 
1401
        if path is not None:
 
1402
            path = path.encode('utf8')
 
1403
        parent_index = self._get_parent_index()
 
1404
        return self._dirstate._get_entry(parent_index, fileid_utf8=file_id, path_utf8=path)
 
1405
 
 
1406
    def _generate_inventory(self):
 
1407
        """Create and set self.inventory from the dirstate object.
 
1408
 
 
1409
        (So this is only called the first time the inventory is requested for
 
1410
        this tree; it then remains in memory until it's out of date.)
 
1411
 
 
1412
        This is relatively expensive: we have to walk the entire dirstate.
 
1413
        """
 
1414
        assert self._locked, 'cannot generate inventory of an unlocked '\
 
1415
            'dirstate revision tree'
 
1416
        # separate call for profiling - makes it clear where the costs are.
 
1417
        self._dirstate._read_dirblocks_if_needed()
 
1418
        assert self._revision_id in self._dirstate.get_parent_ids(), \
 
1419
            'parent %s has disappeared from %s' % (
 
1420
            self._revision_id, self._dirstate.get_parent_ids())
 
1421
        parent_index = self._dirstate.get_parent_ids().index(self._revision_id) + 1
 
1422
        # This is identical now to the WorkingTree _generate_inventory except
 
1423
        # for the tree index use.
 
1424
        root_key, current_entry = self._dirstate._get_entry(parent_index, path_utf8='')
 
1425
        current_id = root_key[2]
 
1426
        assert current_entry[parent_index][0] == 'd'
 
1427
        inv = Inventory(root_id=current_id, revision_id=self._revision_id)
 
1428
        inv.root.revision = current_entry[parent_index][4]
 
1429
        # Turn some things into local variables
 
1430
        minikind_to_kind = dirstate.DirState._minikind_to_kind
 
1431
        factory = entry_factory
 
1432
        utf8_decode = cache_utf8._utf8_decode
 
1433
        inv_byid = inv._byid
 
1434
        # we could do this straight out of the dirstate; it might be fast
 
1435
        # and should be profiled - RBC 20070216
 
1436
        parent_ies = {'' : inv.root}
 
1437
        for block in self._dirstate._dirblocks[1:]: #skip root
 
1438
            dirname = block[0]
 
1439
            try:
 
1440
                parent_ie = parent_ies[dirname]
 
1441
            except KeyError:
 
1442
                # all the paths in this block are not versioned in this tree
 
1443
                continue
 
1444
            for key, entry in block[1]:
 
1445
                minikind, fingerprint, size, executable, revid = entry[parent_index]
 
1446
                if minikind in ('a', 'r'): # absent, relocated
 
1447
                    # not this tree
 
1448
                    continue
 
1449
                name = key[1]
 
1450
                name_unicode = utf8_decode(name)[0]
 
1451
                file_id = key[2]
 
1452
                kind = minikind_to_kind[minikind]
 
1453
                inv_entry = factory[kind](file_id, name_unicode,
 
1454
                                          parent_ie.file_id)
 
1455
                inv_entry.revision = revid
 
1456
                if kind == 'file':
 
1457
                    inv_entry.executable = executable
 
1458
                    inv_entry.text_size = size
 
1459
                    inv_entry.text_sha1 = fingerprint
 
1460
                elif kind == 'directory':
 
1461
                    parent_ies[(dirname + '/' + name).strip('/')] = inv_entry
 
1462
                elif kind == 'symlink':
 
1463
                    inv_entry.executable = False
 
1464
                    inv_entry.text_size = None
 
1465
                    inv_entry.symlink_target = utf8_decode(fingerprint)[0]
 
1466
                elif kind == 'tree-reference':
 
1467
                    inv_entry.reference_revision = fingerprint or None
 
1468
                else:
 
1469
                    raise AssertionError("cannot convert entry %r into an InventoryEntry"
 
1470
                            % entry)
 
1471
                # These checks cost us around 40ms on a 55k entry tree
 
1472
                assert file_id not in inv_byid
 
1473
                assert name_unicode not in parent_ie.children
 
1474
                inv_byid[file_id] = inv_entry
 
1475
                parent_ie.children[name_unicode] = inv_entry
 
1476
        self._inventory = inv
 
1477
 
 
1478
    def get_file_mtime(self, file_id, path=None):
 
1479
        """Return the modification time for this record.
 
1480
 
 
1481
        We return the timestamp of the last-changed revision.
 
1482
        """
 
1483
        # Make sure the file exists
 
1484
        entry = self._get_entry(file_id, path=path)
 
1485
        if entry == (None, None): # do we raise?
 
1486
            return None
 
1487
        parent_index = self._get_parent_index()
 
1488
        last_changed_revision = entry[1][parent_index][4]
 
1489
        return self._repository.get_revision(last_changed_revision).timestamp
 
1490
 
 
1491
    def get_file_sha1(self, file_id, path=None, stat_value=None):
 
1492
        entry = self._get_entry(file_id=file_id, path=path)
 
1493
        parent_index = self._get_parent_index()
 
1494
        parent_details = entry[1][parent_index]
 
1495
        if parent_details[0] == 'f':
 
1496
            return parent_details[1]
 
1497
        return None
 
1498
 
 
1499
    @symbol_versioning.deprecated_method(symbol_versioning.zero_ninety)
 
1500
    def get_weave(self, file_id):
 
1501
        return self._get_weave(file_id)
 
1502
 
 
1503
    def _get_weave(self, file_id):
 
1504
        return self._repository.weave_store.get_weave(file_id,
 
1505
                self._repository.get_transaction())
 
1506
 
 
1507
    def get_file(self, file_id, path=None):
 
1508
        return StringIO(self.get_file_text(file_id))
 
1509
 
 
1510
    def get_file_lines(self, file_id):
 
1511
        ie = self.inventory[file_id]
 
1512
        return self._get_weave(file_id).get_lines(ie.revision)
 
1513
 
 
1514
    def get_file_size(self, file_id):
 
1515
        return self.inventory[file_id].text_size
 
1516
 
 
1517
    def get_file_text(self, file_id):
 
1518
        return ''.join(self.get_file_lines(file_id))
 
1519
 
 
1520
    def get_reference_revision(self, file_id, path=None):
 
1521
        return self.inventory[file_id].reference_revision
 
1522
 
 
1523
    def iter_files_bytes(self, desired_files):
 
1524
        """See Tree.iter_files_bytes.
 
1525
 
 
1526
        This version is implemented on top of Repository.iter_files_bytes"""
 
1527
        parent_index = self._get_parent_index()
 
1528
        repo_desired_files = []
 
1529
        for file_id, identifier in desired_files:
 
1530
            entry = self._get_entry(file_id)
 
1531
            if entry == (None, None):
 
1532
                raise errors.NoSuchId(self, file_id)
 
1533
            repo_desired_files.append((file_id, entry[1][parent_index][4],
 
1534
                                       identifier))
 
1535
        return self._repository.iter_files_bytes(repo_desired_files)
 
1536
 
 
1537
    def get_symlink_target(self, file_id):
 
1538
        entry = self._get_entry(file_id=file_id)
 
1539
        parent_index = self._get_parent_index()
 
1540
        if entry[1][parent_index][0] != 'l':
 
1541
            return None
 
1542
        else:
 
1543
            # At present, none of the tree implementations supports non-ascii
 
1544
            # symlink targets. So we will just assume that the dirstate path is
 
1545
            # correct.
 
1546
            return entry[1][parent_index][1]
 
1547
 
 
1548
    def get_revision_id(self):
 
1549
        """Return the revision id for this tree."""
 
1550
        return self._revision_id
 
1551
 
 
1552
    def _get_inventory(self):
 
1553
        if self._inventory is not None:
 
1554
            return self._inventory
 
1555
        self._must_be_locked()
 
1556
        self._generate_inventory()
 
1557
        return self._inventory
 
1558
 
 
1559
    inventory = property(_get_inventory,
 
1560
                         doc="Inventory of this Tree")
 
1561
 
 
1562
    def get_parent_ids(self):
 
1563
        """The parents of a tree in the dirstate are not cached."""
 
1564
        return self._repository.get_revision(self._revision_id).parent_ids
 
1565
 
 
1566
    def has_filename(self, filename):
 
1567
        return bool(self.path2id(filename))
 
1568
 
 
1569
    def kind(self, file_id):
 
1570
        return self.inventory[file_id].kind
 
1571
 
 
1572
    def is_executable(self, file_id, path=None):
 
1573
        ie = self.inventory[file_id]
 
1574
        if ie.kind != "file":
 
1575
            return None
 
1576
        return ie.executable
 
1577
 
 
1578
    def list_files(self, include_root=False):
 
1579
        # We use a standard implementation, because DirStateRevisionTree is
 
1580
        # dealing with one of the parents of the current state
 
1581
        inv = self._get_inventory()
 
1582
        entries = inv.iter_entries()
 
1583
        if self.inventory.root is not None and not include_root:
 
1584
            entries.next()
 
1585
        for path, entry in entries:
 
1586
            yield path, 'V', entry.kind, entry.file_id, entry
 
1587
 
 
1588
    def lock_read(self):
 
1589
        """Lock the tree for a set of operations."""
 
1590
        if not self._locked:
 
1591
            self._repository.lock_read()
 
1592
            if self._dirstate._lock_token is None:
 
1593
                self._dirstate.lock_read()
 
1594
                self._dirstate_locked = True
 
1595
        self._locked += 1
 
1596
 
 
1597
    def _must_be_locked(self):
 
1598
        if not self._locked:
 
1599
            raise errors.ObjectNotLocked(self)
 
1600
 
 
1601
    @needs_read_lock
 
1602
    def path2id(self, path):
 
1603
        """Return the id for path in this tree."""
 
1604
        # lookup by path: faster than splitting and walking the ivnentory.
 
1605
        entry = self._get_entry(path=path)
 
1606
        if entry == (None, None):
 
1607
            return None
 
1608
        return entry[0][2]
 
1609
 
 
1610
    def unlock(self):
 
1611
        """Unlock, freeing any cache memory used during the lock."""
 
1612
        # outside of a lock, the inventory is suspect: release it.
 
1613
        self._locked -=1
 
1614
        if not self._locked:
 
1615
            self._inventory = None
 
1616
            self._locked = 0
 
1617
            if self._dirstate_locked:
 
1618
                self._dirstate.unlock()
 
1619
                self._dirstate_locked = False
 
1620
            self._repository.unlock()
 
1621
 
 
1622
    def walkdirs(self, prefix=""):
 
1623
        # TODO: jam 20070215 This is the lazy way by using the RevisionTree
 
1624
        # implementation based on an inventory.  
 
1625
        # This should be cleaned up to use the much faster Dirstate code
 
1626
        # So for now, we just build up the parent inventory, and extract
 
1627
        # it the same way RevisionTree does.
 
1628
        _directory = 'directory'
 
1629
        inv = self._get_inventory()
 
1630
        top_id = inv.path2id(prefix)
 
1631
        if top_id is None:
 
1632
            pending = []
 
1633
        else:
 
1634
            pending = [(prefix, top_id)]
 
1635
        while pending:
 
1636
            dirblock = []
 
1637
            relpath, file_id = pending.pop()
 
1638
            # 0 - relpath, 1- file-id
 
1639
            if relpath:
 
1640
                relroot = relpath + '/'
 
1641
            else:
 
1642
                relroot = ""
 
1643
            # FIXME: stash the node in pending
 
1644
            entry = inv[file_id]
 
1645
            for name, child in entry.sorted_children():
 
1646
                toppath = relroot + name
 
1647
                dirblock.append((toppath, name, child.kind, None,
 
1648
                    child.file_id, child.kind
 
1649
                    ))
 
1650
            yield (relpath, entry.file_id), dirblock
 
1651
            # push the user specified dirs from dirblock
 
1652
            for dir in reversed(dirblock):
 
1653
                if dir[2] == _directory:
 
1654
                    pending.append((dir[0], dir[4]))
 
1655
 
 
1656
 
 
1657
class InterDirStateTree(InterTree):
 
1658
    """Fast path optimiser for changes_from with dirstate trees.
 
1659
    
 
1660
    This is used only when both trees are in the dirstate working file, and 
 
1661
    the source is any parent within the dirstate, and the destination is 
 
1662
    the current working tree of the same dirstate.
 
1663
    """
 
1664
    # this could be generalized to allow comparisons between any trees in the
 
1665
    # dirstate, and possibly between trees stored in different dirstates.
 
1666
 
 
1667
    def __init__(self, source, target):
 
1668
        super(InterDirStateTree, self).__init__(source, target)
 
1669
        if not InterDirStateTree.is_compatible(source, target):
 
1670
            raise Exception, "invalid source %r and target %r" % (source, target)
 
1671
 
 
1672
    @staticmethod
 
1673
    def make_source_parent_tree(source, target):
 
1674
        """Change the source tree into a parent of the target."""
 
1675
        revid = source.commit('record tree')
 
1676
        target.branch.repository.fetch(source.branch.repository, revid)
 
1677
        target.set_parent_ids([revid])
 
1678
        return target.basis_tree(), target
 
1679
 
 
1680
    _matching_from_tree_format = WorkingTreeFormat4()
 
1681
    _matching_to_tree_format = WorkingTreeFormat4()
 
1682
    _test_mutable_trees_to_test_trees = make_source_parent_tree
 
1683
 
 
1684
    def _iter_changes(self, include_unchanged=False,
 
1685
                      specific_files=None, pb=None, extra_trees=[],
 
1686
                      require_versioned=True, want_unversioned=False):
 
1687
        """Return the changes from source to target.
 
1688
 
 
1689
        :return: An iterator that yields tuples. See InterTree._iter_changes
 
1690
            for details.
 
1691
        :param specific_files: An optional list of file paths to restrict the
 
1692
            comparison to. When mapping filenames to ids, all matches in all
 
1693
            trees (including optional extra_trees) are used, and all children of
 
1694
            matched directories are included.
 
1695
        :param include_unchanged: An optional boolean requesting the inclusion of
 
1696
            unchanged entries in the result.
 
1697
        :param extra_trees: An optional list of additional trees to use when
 
1698
            mapping the contents of specific_files (paths) to file_ids.
 
1699
        :param require_versioned: If True, all files in specific_files must be
 
1700
            versioned in one of source, target, extra_trees or
 
1701
            PathsNotVersionedError is raised.
 
1702
        :param want_unversioned: Should unversioned files be returned in the
 
1703
            output. An unversioned file is defined as one with (False, False)
 
1704
            for the versioned pair.
 
1705
        """
 
1706
        utf8_decode = cache_utf8._utf8_decode
 
1707
        _minikind_to_kind = dirstate.DirState._minikind_to_kind
 
1708
        cmp_by_dirs = dirstate.cmp_by_dirs
 
1709
        # NB: show_status depends on being able to pass in non-versioned files
 
1710
        # and report them as unknown
 
1711
        # TODO: handle extra trees in the dirstate.
 
1712
        # TODO: handle comparisons as an empty tree as a different special
 
1713
        # case? mbp 20070226
 
1714
        if (extra_trees or (self.source._revision_id == NULL_REVISION)
 
1715
            or specific_files == []):
 
1716
            # we can't fast-path these cases (yet)
 
1717
            for f in super(InterDirStateTree, self)._iter_changes(
 
1718
                include_unchanged, specific_files, pb, extra_trees,
 
1719
                require_versioned, want_unversioned=want_unversioned):
 
1720
                yield f
 
1721
            return
 
1722
        parent_ids = self.target.get_parent_ids()
 
1723
        assert (self.source._revision_id in parent_ids), \
 
1724
                "revision {%s} is not stored in {%s}, but %s " \
 
1725
                "can only be used for trees stored in the dirstate" \
 
1726
                % (self.source._revision_id, self.target, self._iter_changes)
 
1727
        target_index = 0
 
1728
        if self.source._revision_id == NULL_REVISION:
 
1729
            source_index = None
 
1730
            indices = (target_index,)
 
1731
        else:
 
1732
            assert (self.source._revision_id in parent_ids), \
 
1733
                "Failure: source._revision_id: %s not in target.parent_ids(%s)" % (
 
1734
                self.source._revision_id, parent_ids)
 
1735
            source_index = 1 + parent_ids.index(self.source._revision_id)
 
1736
            indices = (source_index,target_index)
 
1737
        # -- make all specific_files utf8 --
 
1738
        if specific_files:
 
1739
            specific_files_utf8 = set()
 
1740
            for path in specific_files:
 
1741
                specific_files_utf8.add(path.encode('utf8'))
 
1742
            specific_files = specific_files_utf8
 
1743
        else:
 
1744
            specific_files = set([''])
 
1745
        # -- specific_files is now a utf8 path set --
 
1746
        # -- get the state object and prepare it.
 
1747
        state = self.target.current_dirstate()
 
1748
        state._read_dirblocks_if_needed()
 
1749
        def _entries_for_path(path):
 
1750
            """Return a list with all the entries that match path for all ids.
 
1751
            """
 
1752
            dirname, basename = os.path.split(path)
 
1753
            key = (dirname, basename, '')
 
1754
            block_index, present = state._find_block_index_from_key(key)
 
1755
            if not present:
 
1756
                # the block which should contain path is absent.
 
1757
                return []
 
1758
            result = []
 
1759
            block = state._dirblocks[block_index][1]
 
1760
            entry_index, _ = state._find_entry_index(key, block)
 
1761
            # we may need to look at multiple entries at this path: walk while the specific_files match.
 
1762
            while (entry_index < len(block) and
 
1763
                block[entry_index][0][0:2] == key[0:2]):
 
1764
                result.append(block[entry_index])
 
1765
                entry_index += 1
 
1766
            return result
 
1767
        if require_versioned:
 
1768
            # -- check all supplied paths are versioned in a search tree. --
 
1769
            all_versioned = True
 
1770
            for path in specific_files:
 
1771
                path_entries = _entries_for_path(path)
 
1772
                if not path_entries:
 
1773
                    # this specified path is not present at all: error
 
1774
                    all_versioned = False
 
1775
                    break
 
1776
                found_versioned = False
 
1777
                # for each id at this path
 
1778
                for entry in path_entries:
 
1779
                    # for each tree.
 
1780
                    for index in indices:
 
1781
                        if entry[1][index][0] != 'a': # absent
 
1782
                            found_versioned = True
 
1783
                            # all good: found a versioned cell
 
1784
                            break
 
1785
                if not found_versioned:
 
1786
                    # none of the indexes was not 'absent' at all ids for this
 
1787
                    # path.
 
1788
                    all_versioned = False
 
1789
                    break
 
1790
            if not all_versioned:
 
1791
                raise errors.PathsNotVersionedError(specific_files)
 
1792
        # -- remove redundancy in supplied specific_files to prevent over-scanning --
 
1793
        search_specific_files = set()
 
1794
        for path in specific_files:
 
1795
            other_specific_files = specific_files.difference(set([path]))
 
1796
            if not osutils.is_inside_any(other_specific_files, path):
 
1797
                # this is a top level path, we must check it.
 
1798
                search_specific_files.add(path)
 
1799
        # sketch: 
 
1800
        # compare source_index and target_index at or under each element of search_specific_files.
 
1801
        # follow the following comparison table. Note that we only want to do diff operations when
 
1802
        # the target is fdl because thats when the walkdirs logic will have exposed the pathinfo 
 
1803
        # for the target.
 
1804
        # cases:
 
1805
        # 
 
1806
        # Source | Target | disk | action
 
1807
        #   r    | fdlt   |      | add source to search, add id path move and perform
 
1808
        #        |        |      | diff check on source-target
 
1809
        #   r    | fdlt   |  a   | dangling file that was present in the basis. 
 
1810
        #        |        |      | ???
 
1811
        #   r    |  a     |      | add source to search
 
1812
        #   r    |  a     |  a   | 
 
1813
        #   r    |  r     |      | this path is present in a non-examined tree, skip.
 
1814
        #   r    |  r     |  a   | this path is present in a non-examined tree, skip.
 
1815
        #   a    | fdlt   |      | add new id
 
1816
        #   a    | fdlt   |  a   | dangling locally added file, skip
 
1817
        #   a    |  a     |      | not present in either tree, skip
 
1818
        #   a    |  a     |  a   | not present in any tree, skip
 
1819
        #   a    |  r     |      | not present in either tree at this path, skip as it
 
1820
        #        |        |      | may not be selected by the users list of paths.
 
1821
        #   a    |  r     |  a   | not present in either tree at this path, skip as it
 
1822
        #        |        |      | may not be selected by the users list of paths.
 
1823
        #  fdlt  | fdlt   |      | content in both: diff them
 
1824
        #  fdlt  | fdlt   |  a   | deleted locally, but not unversioned - show as deleted ?
 
1825
        #  fdlt  |  a     |      | unversioned: output deleted id for now
 
1826
        #  fdlt  |  a     |  a   | unversioned and deleted: output deleted id
 
1827
        #  fdlt  |  r     |      | relocated in this tree, so add target to search.
 
1828
        #        |        |      | Dont diff, we will see an r,fd; pair when we reach
 
1829
        #        |        |      | this id at the other path.
 
1830
        #  fdlt  |  r     |  a   | relocated in this tree, so add target to search.
 
1831
        #        |        |      | Dont diff, we will see an r,fd; pair when we reach
 
1832
        #        |        |      | this id at the other path.
 
1833
 
 
1834
        # for all search_indexs in each path at or under each element of
 
1835
        # search_specific_files, if the detail is relocated: add the id, and add the
 
1836
        # relocated path as one to search if its not searched already. If the
 
1837
        # detail is not relocated, add the id.
 
1838
        searched_specific_files = set()
 
1839
        NULL_PARENT_DETAILS = dirstate.DirState.NULL_PARENT_DETAILS
 
1840
        # Using a list so that we can access the values and change them in
 
1841
        # nested scope. Each one is [path, file_id, entry]
 
1842
        last_source_parent = [None, None]
 
1843
        last_target_parent = [None, None]
 
1844
 
 
1845
        use_filesystem_for_exec = (sys.platform != 'win32')
 
1846
 
 
1847
        # Just a sentry, so that _process_entry can say that this
 
1848
        # record is handled, but isn't interesting to process (unchanged)
 
1849
        uninteresting = object()
 
1850
 
 
1851
 
 
1852
        old_dirname_to_file_id = {}
 
1853
        new_dirname_to_file_id = {}
 
1854
        # TODO: jam 20070516 - Avoid the _get_entry lookup overhead by
 
1855
        #       keeping a cache of directories that we have seen.
 
1856
 
 
1857
        def _process_entry(entry, path_info):
 
1858
            """Compare an entry and real disk to generate delta information.
 
1859
 
 
1860
            :param path_info: top_relpath, basename, kind, lstat, abspath for
 
1861
                the path of entry. If None, then the path is considered absent.
 
1862
                (Perhaps we should pass in a concrete entry for this ?)
 
1863
                Basename is returned as a utf8 string because we expect this
 
1864
                tuple will be ignored, and don't want to take the time to
 
1865
                decode.
 
1866
            :return: None if these don't match
 
1867
                     A tuple of information about the change, or
 
1868
                     the object 'uninteresting' if these match, but are
 
1869
                     basically identical.
 
1870
            """
 
1871
            if source_index is None:
 
1872
                source_details = NULL_PARENT_DETAILS
 
1873
            else:
 
1874
                source_details = entry[1][source_index]
 
1875
            target_details = entry[1][target_index]
 
1876
            target_minikind = target_details[0]
 
1877
            if path_info is not None and target_minikind in 'fdlt':
 
1878
                assert target_index == 0
 
1879
                link_or_sha1 = state.update_entry(entry, abspath=path_info[4],
 
1880
                                                  stat_value=path_info[3])
 
1881
                # The entry may have been modified by update_entry
 
1882
                target_details = entry[1][target_index]
 
1883
                target_minikind = target_details[0]
 
1884
            else:
 
1885
                link_or_sha1 = None
 
1886
            file_id = entry[0][2]
 
1887
            source_minikind = source_details[0]
 
1888
            if source_minikind in 'fdltr' and target_minikind in 'fdlt':
 
1889
                # claimed content in both: diff
 
1890
                #   r    | fdlt   |      | add source to search, add id path move and perform
 
1891
                #        |        |      | diff check on source-target
 
1892
                #   r    | fdlt   |  a   | dangling file that was present in the basis.
 
1893
                #        |        |      | ???
 
1894
                if source_minikind in 'r':
 
1895
                    # add the source to the search path to find any children it
 
1896
                    # has.  TODO ? : only add if it is a container ?
 
1897
                    if not osutils.is_inside_any(searched_specific_files,
 
1898
                                                 source_details[1]):
 
1899
                        search_specific_files.add(source_details[1])
 
1900
                    # generate the old path; this is needed for stating later
 
1901
                    # as well.
 
1902
                    old_path = source_details[1]
 
1903
                    old_dirname, old_basename = os.path.split(old_path)
 
1904
                    path = pathjoin(entry[0][0], entry[0][1])
 
1905
                    old_entry = state._get_entry(source_index,
 
1906
                                                 path_utf8=old_path)
 
1907
                    # update the source details variable to be the real
 
1908
                    # location.
 
1909
                    source_details = old_entry[1][source_index]
 
1910
                    source_minikind = source_details[0]
 
1911
                else:
 
1912
                    old_dirname = entry[0][0]
 
1913
                    old_basename = entry[0][1]
 
1914
                    old_path = path = None
 
1915
                if path_info is None:
 
1916
                    # the file is missing on disk, show as removed.
 
1917
                    content_change = True
 
1918
                    target_kind = None
 
1919
                    target_exec = False
 
1920
                else:
 
1921
                    # source and target are both versioned and disk file is present.
 
1922
                    target_kind = path_info[2]
 
1923
                    if target_kind == 'directory':
 
1924
                        if path is None:
 
1925
                            old_path = path = pathjoin(old_dirname, old_basename)
 
1926
                        new_dirname_to_file_id[path] = file_id
 
1927
                        if source_minikind != 'd':
 
1928
                            content_change = True
 
1929
                        else:
 
1930
                            # directories have no fingerprint
 
1931
                            content_change = False
 
1932
                        target_exec = False
 
1933
                    elif target_kind == 'file':
 
1934
                        if source_minikind != 'f':
 
1935
                            content_change = True
 
1936
                        else:
 
1937
                            # We could check the size, but we already have the
 
1938
                            # sha1 hash.
 
1939
                            content_change = (link_or_sha1 != source_details[1])
 
1940
                        # Target details is updated at update_entry time
 
1941
                        if use_filesystem_for_exec:
 
1942
                            # We don't need S_ISREG here, because we are sure
 
1943
                            # we are dealing with a file.
 
1944
                            target_exec = bool(stat.S_IEXEC & path_info[3].st_mode)
 
1945
                        else:
 
1946
                            target_exec = target_details[3]
 
1947
                    elif target_kind == 'symlink':
 
1948
                        if source_minikind != 'l':
 
1949
                            content_change = True
 
1950
                        else:
 
1951
                            content_change = (link_or_sha1 != source_details[1])
 
1952
                        target_exec = False
 
1953
                    elif target_kind == 'tree-reference':
 
1954
                        if source_minikind != 't':
 
1955
                            content_change = True
 
1956
                        else:
 
1957
                            content_change = False
 
1958
                        target_exec = False
 
1959
                    else:
 
1960
                        raise Exception, "unknown kind %s" % path_info[2]
 
1961
                if source_minikind == 'd':
 
1962
                    if path is None:
 
1963
                        old_path = path = pathjoin(old_dirname, old_basename)
 
1964
                    old_dirname_to_file_id[old_path] = file_id
 
1965
                # parent id is the entry for the path in the target tree
 
1966
                if old_dirname == last_source_parent[0]:
 
1967
                    source_parent_id = last_source_parent[1]
 
1968
                else:
 
1969
                    try:
 
1970
                        source_parent_id = old_dirname_to_file_id[old_dirname]
 
1971
                    except KeyError:
 
1972
                        source_parent_entry = state._get_entry(source_index,
 
1973
                                                               path_utf8=old_dirname)
 
1974
                        source_parent_id = source_parent_entry[0][2]
 
1975
                    if source_parent_id == entry[0][2]:
 
1976
                        # This is the root, so the parent is None
 
1977
                        source_parent_id = None
 
1978
                    else:
 
1979
                        last_source_parent[0] = old_dirname
 
1980
                        last_source_parent[1] = source_parent_id
 
1981
                new_dirname = entry[0][0]
 
1982
                if new_dirname == last_target_parent[0]:
 
1983
                    target_parent_id = last_target_parent[1]
 
1984
                else:
 
1985
                    try:
 
1986
                        target_parent_id = new_dirname_to_file_id[new_dirname]
 
1987
                    except KeyError:
 
1988
                        # TODO: We don't always need to do the lookup, because the
 
1989
                        #       parent entry will be the same as the source entry.
 
1990
                        target_parent_entry = state._get_entry(target_index,
 
1991
                                                               path_utf8=new_dirname)
 
1992
                        assert target_parent_entry != (None, None), (
 
1993
                            "Could not find target parent in wt: %s\nparent of: %s"
 
1994
                            % (new_dirname, entry))
 
1995
                        target_parent_id = target_parent_entry[0][2]
 
1996
                    if target_parent_id == entry[0][2]:
 
1997
                        # This is the root, so the parent is None
 
1998
                        target_parent_id = None
 
1999
                    else:
 
2000
                        last_target_parent[0] = new_dirname
 
2001
                        last_target_parent[1] = target_parent_id
 
2002
 
 
2003
                source_exec = source_details[3]
 
2004
                if (include_unchanged
 
2005
                    or content_change
 
2006
                    or source_parent_id != target_parent_id
 
2007
                    or old_basename != entry[0][1]
 
2008
                    or source_exec != target_exec
 
2009
                    ):
 
2010
                    if old_path is None:
 
2011
                        old_path = path = pathjoin(old_dirname, old_basename)
 
2012
                        old_path_u = utf8_decode(old_path)[0]
 
2013
                        path_u = old_path_u
 
2014
                    else:
 
2015
                        old_path_u = utf8_decode(old_path)[0]
 
2016
                        if old_path == path:
 
2017
                            path_u = old_path_u
 
2018
                        else:
 
2019
                            path_u = utf8_decode(path)[0]
 
2020
                    source_kind = _minikind_to_kind[source_minikind]
 
2021
                    return (entry[0][2],
 
2022
                           (old_path_u, path_u),
 
2023
                           content_change,
 
2024
                           (True, True),
 
2025
                           (source_parent_id, target_parent_id),
 
2026
                           (utf8_decode(old_basename)[0], utf8_decode(entry[0][1])[0]),
 
2027
                           (source_kind, target_kind),
 
2028
                           (source_exec, target_exec))
 
2029
                else:
 
2030
                    return uninteresting
 
2031
            elif source_minikind in 'a' and target_minikind in 'fdlt':
 
2032
                # looks like a new file
 
2033
                if path_info is not None:
 
2034
                    path = pathjoin(entry[0][0], entry[0][1])
 
2035
                    # parent id is the entry for the path in the target tree
 
2036
                    # TODO: these are the same for an entire directory: cache em.
 
2037
                    parent_id = state._get_entry(target_index,
 
2038
                                                 path_utf8=entry[0][0])[0][2]
 
2039
                    if parent_id == entry[0][2]:
 
2040
                        parent_id = None
 
2041
                    if use_filesystem_for_exec:
 
2042
                        # We need S_ISREG here, because we aren't sure if this
 
2043
                        # is a file or not.
 
2044
                        target_exec = bool(
 
2045
                            stat.S_ISREG(path_info[3].st_mode)
 
2046
                            and stat.S_IEXEC & path_info[3].st_mode)
 
2047
                    else:
 
2048
                        target_exec = target_details[3]
 
2049
                    return (entry[0][2],
 
2050
                           (None, utf8_decode(path)[0]),
 
2051
                           True,
 
2052
                           (False, True),
 
2053
                           (None, parent_id),
 
2054
                           (None, utf8_decode(entry[0][1])[0]),
 
2055
                           (None, path_info[2]),
 
2056
                           (None, target_exec))
 
2057
                else:
 
2058
                    # but its not on disk: we deliberately treat this as just
 
2059
                    # never-present. (Why ?! - RBC 20070224)
 
2060
                    pass
 
2061
            elif source_minikind in 'fdlt' and target_minikind in 'a':
 
2062
                # unversioned, possibly, or possibly not deleted: we dont care.
 
2063
                # if its still on disk, *and* theres no other entry at this
 
2064
                # path [we dont know this in this routine at the moment -
 
2065
                # perhaps we should change this - then it would be an unknown.
 
2066
                old_path = pathjoin(entry[0][0], entry[0][1])
 
2067
                # parent id is the entry for the path in the target tree
 
2068
                parent_id = state._get_entry(source_index, path_utf8=entry[0][0])[0][2]
 
2069
                if parent_id == entry[0][2]:
 
2070
                    parent_id = None
 
2071
                return (entry[0][2],
 
2072
                       (utf8_decode(old_path)[0], None),
 
2073
                       True,
 
2074
                       (True, False),
 
2075
                       (parent_id, None),
 
2076
                       (utf8_decode(entry[0][1])[0], None),
 
2077
                       (_minikind_to_kind[source_minikind], None),
 
2078
                       (source_details[3], None))
 
2079
            elif source_minikind in 'fdlt' and target_minikind in 'r':
 
2080
                # a rename; could be a true rename, or a rename inherited from
 
2081
                # a renamed parent. TODO: handle this efficiently. Its not
 
2082
                # common case to rename dirs though, so a correct but slow
 
2083
                # implementation will do.
 
2084
                if not osutils.is_inside_any(searched_specific_files, target_details[1]):
 
2085
                    search_specific_files.add(target_details[1])
 
2086
            elif source_minikind in 'ra' and target_minikind in 'ra':
 
2087
                # neither of the selected trees contain this file,
 
2088
                # so skip over it. This is not currently directly tested, but
 
2089
                # is indirectly via test_too_much.TestCommands.test_conflicts.
 
2090
                pass
 
2091
            else:
 
2092
                raise AssertionError("don't know how to compare "
 
2093
                    "source_minikind=%r, target_minikind=%r"
 
2094
                    % (source_minikind, target_minikind))
 
2095
                ## import pdb;pdb.set_trace()
 
2096
            return None
 
2097
 
 
2098
        while search_specific_files:
 
2099
            # TODO: the pending list should be lexically sorted?  the
 
2100
            # interface doesn't require it.
 
2101
            current_root = search_specific_files.pop()
 
2102
            current_root_unicode = current_root.decode('utf8')
 
2103
            searched_specific_files.add(current_root)
 
2104
            # process the entries for this containing directory: the rest will be
 
2105
            # found by their parents recursively.
 
2106
            root_entries = _entries_for_path(current_root)
 
2107
            root_abspath = self.target.abspath(current_root_unicode)
 
2108
            try:
 
2109
                root_stat = os.lstat(root_abspath)
 
2110
            except OSError, e:
 
2111
                if e.errno == errno.ENOENT:
 
2112
                    # the path does not exist: let _process_entry know that.
 
2113
                    root_dir_info = None
 
2114
                else:
 
2115
                    # some other random error: hand it up.
 
2116
                    raise
 
2117
            else:
 
2118
                root_dir_info = ('', current_root,
 
2119
                    osutils.file_kind_from_stat_mode(root_stat.st_mode), root_stat,
 
2120
                    root_abspath)
 
2121
                if root_dir_info[2] == 'directory':
 
2122
                    if self.target._directory_is_tree_reference(
 
2123
                        current_root.decode('utf8')):
 
2124
                        root_dir_info = root_dir_info[:2] + \
 
2125
                            ('tree-reference',) + root_dir_info[3:]
 
2126
 
 
2127
            if not root_entries and not root_dir_info:
 
2128
                # this specified path is not present at all, skip it.
 
2129
                continue
 
2130
            path_handled = False
 
2131
            for entry in root_entries:
 
2132
                result = _process_entry(entry, root_dir_info)
 
2133
                if result is not None:
 
2134
                    path_handled = True
 
2135
                    if result is not uninteresting:
 
2136
                        yield result
 
2137
            if want_unversioned and not path_handled and root_dir_info:
 
2138
                new_executable = bool(
 
2139
                    stat.S_ISREG(root_dir_info[3].st_mode)
 
2140
                    and stat.S_IEXEC & root_dir_info[3].st_mode)
 
2141
                yield (None,
 
2142
                       (None, current_root_unicode),
 
2143
                       True,
 
2144
                       (False, False),
 
2145
                       (None, None),
 
2146
                       (None, splitpath(current_root_unicode)[-1]),
 
2147
                       (None, root_dir_info[2]),
 
2148
                       (None, new_executable)
 
2149
                      )
 
2150
            initial_key = (current_root, '', '')
 
2151
            block_index, _ = state._find_block_index_from_key(initial_key)
 
2152
            if block_index == 0:
 
2153
                # we have processed the total root already, but because the
 
2154
                # initial key matched it we should skip it here.
 
2155
                block_index +=1
 
2156
            if root_dir_info and root_dir_info[2] == 'tree-reference':
 
2157
                current_dir_info = None
 
2158
            else:
 
2159
                dir_iterator = osutils._walkdirs_utf8(root_abspath, prefix=current_root)
 
2160
                try:
 
2161
                    current_dir_info = dir_iterator.next()
 
2162
                except OSError, e:
 
2163
                    # on win32, python2.4 has e.errno == ERROR_DIRECTORY, but
 
2164
                    # python 2.5 has e.errno == EINVAL,
 
2165
                    #            and e.winerror == ERROR_DIRECTORY
 
2166
                    e_winerror = getattr(e, 'winerror', None)
 
2167
                    win_errors = (ERROR_DIRECTORY, ERROR_PATH_NOT_FOUND)
 
2168
                    # there may be directories in the inventory even though
 
2169
                    # this path is not a file on disk: so mark it as end of
 
2170
                    # iterator
 
2171
                    if e.errno in (errno.ENOENT, errno.ENOTDIR, errno.EINVAL):
 
2172
                        current_dir_info = None
 
2173
                    elif (sys.platform == 'win32'
 
2174
                          and (e.errno in win_errors
 
2175
                               or e_winerror in win_errors)):
 
2176
                        current_dir_info = None
 
2177
                    else:
 
2178
                        raise
 
2179
                else:
 
2180
                    if current_dir_info[0][0] == '':
 
2181
                        # remove .bzr from iteration
 
2182
                        bzr_index = bisect_left(current_dir_info[1], ('.bzr',))
 
2183
                        assert current_dir_info[1][bzr_index][0] == '.bzr'
 
2184
                        del current_dir_info[1][bzr_index]
 
2185
            # walk until both the directory listing and the versioned metadata
 
2186
            # are exhausted. 
 
2187
            if (block_index < len(state._dirblocks) and
 
2188
                osutils.is_inside(current_root, state._dirblocks[block_index][0])):
 
2189
                current_block = state._dirblocks[block_index]
 
2190
            else:
 
2191
                current_block = None
 
2192
            while (current_dir_info is not None or
 
2193
                   current_block is not None):
 
2194
                if (current_dir_info and current_block
 
2195
                    and current_dir_info[0][0] != current_block[0]):
 
2196
                    if cmp_by_dirs(current_dir_info[0][0], current_block[0]) < 0:
 
2197
                        # filesystem data refers to paths not covered by the dirblock.
 
2198
                        # this has two possibilities:
 
2199
                        # A) it is versioned but empty, so there is no block for it
 
2200
                        # B) it is not versioned.
 
2201
 
 
2202
                        # if (A) then we need to recurse into it to check for
 
2203
                        # new unknown files or directories.
 
2204
                        # if (B) then we should ignore it, because we don't
 
2205
                        # recurse into unknown directories.
 
2206
                        path_index = 0
 
2207
                        while path_index < len(current_dir_info[1]):
 
2208
                                current_path_info = current_dir_info[1][path_index]
 
2209
                                if want_unversioned:
 
2210
                                    if current_path_info[2] == 'directory':
 
2211
                                        if self.target._directory_is_tree_reference(
 
2212
                                            current_path_info[0].decode('utf8')):
 
2213
                                            current_path_info = current_path_info[:2] + \
 
2214
                                                ('tree-reference',) + current_path_info[3:]
 
2215
                                    new_executable = bool(
 
2216
                                        stat.S_ISREG(current_path_info[3].st_mode)
 
2217
                                        and stat.S_IEXEC & current_path_info[3].st_mode)
 
2218
                                    yield (None,
 
2219
                                        (None, utf8_decode(current_path_info[0])[0]),
 
2220
                                        True,
 
2221
                                        (False, False),
 
2222
                                        (None, None),
 
2223
                                        (None, utf8_decode(current_path_info[1])[0]),
 
2224
                                        (None, current_path_info[2]),
 
2225
                                        (None, new_executable))
 
2226
                                # dont descend into this unversioned path if it is
 
2227
                                # a dir
 
2228
                                if current_path_info[2] in ('directory',
 
2229
                                                            'tree-reference'):
 
2230
                                    del current_dir_info[1][path_index]
 
2231
                                    path_index -= 1
 
2232
                                path_index += 1
 
2233
 
 
2234
                        # This dir info has been handled, go to the next
 
2235
                        try:
 
2236
                            current_dir_info = dir_iterator.next()
 
2237
                        except StopIteration:
 
2238
                            current_dir_info = None
 
2239
                    else:
 
2240
                        # We have a dirblock entry for this location, but there
 
2241
                        # is no filesystem path for this. This is most likely
 
2242
                        # because a directory was removed from the disk.
 
2243
                        # We don't have to report the missing directory,
 
2244
                        # because that should have already been handled, but we
 
2245
                        # need to handle all of the files that are contained
 
2246
                        # within.
 
2247
                        for current_entry in current_block[1]:
 
2248
                            # entry referring to file not present on disk.
 
2249
                            # advance the entry only, after processing.
 
2250
                            result = _process_entry(current_entry, None)
 
2251
                            if result is not None:
 
2252
                                if result is not uninteresting:
 
2253
                                    yield result
 
2254
                        block_index +=1
 
2255
                        if (block_index < len(state._dirblocks) and
 
2256
                            osutils.is_inside(current_root,
 
2257
                                              state._dirblocks[block_index][0])):
 
2258
                            current_block = state._dirblocks[block_index]
 
2259
                        else:
 
2260
                            current_block = None
 
2261
                    continue
 
2262
                entry_index = 0
 
2263
                if current_block and entry_index < len(current_block[1]):
 
2264
                    current_entry = current_block[1][entry_index]
 
2265
                else:
 
2266
                    current_entry = None
 
2267
                advance_entry = True
 
2268
                path_index = 0
 
2269
                if current_dir_info and path_index < len(current_dir_info[1]):
 
2270
                    current_path_info = current_dir_info[1][path_index]
 
2271
                    if current_path_info[2] == 'directory':
 
2272
                        if self.target._directory_is_tree_reference(
 
2273
                            current_path_info[0].decode('utf8')):
 
2274
                            current_path_info = current_path_info[:2] + \
 
2275
                                ('tree-reference',) + current_path_info[3:]
 
2276
                else:
 
2277
                    current_path_info = None
 
2278
                advance_path = True
 
2279
                path_handled = False
 
2280
                while (current_entry is not None or
 
2281
                    current_path_info is not None):
 
2282
                    if current_entry is None:
 
2283
                        # the check for path_handled when the path is adnvaced
 
2284
                        # will yield this path if needed.
 
2285
                        pass
 
2286
                    elif current_path_info is None:
 
2287
                        # no path is fine: the per entry code will handle it.
 
2288
                        result = _process_entry(current_entry, current_path_info)
 
2289
                        if result is not None:
 
2290
                            if result is not uninteresting:
 
2291
                                yield result
 
2292
                    elif (current_entry[0][1] != current_path_info[1]
 
2293
                          or current_entry[1][target_index][0] in 'ar'):
 
2294
                        # The current path on disk doesn't match the dirblock
 
2295
                        # record. Either the dirblock is marked as absent, or
 
2296
                        # the file on disk is not present at all in the
 
2297
                        # dirblock. Either way, report about the dirblock
 
2298
                        # entry, and let other code handle the filesystem one.
 
2299
 
 
2300
                        # Compare the basename for these files to determine
 
2301
                        # which comes first
 
2302
                        if current_path_info[1] < current_entry[0][1]:
 
2303
                            # extra file on disk: pass for now, but only
 
2304
                            # increment the path, not the entry
 
2305
                            advance_entry = False
 
2306
                        else:
 
2307
                            # entry referring to file not present on disk.
 
2308
                            # advance the entry only, after processing.
 
2309
                            result = _process_entry(current_entry, None)
 
2310
                            if result is not None:
 
2311
                                if result is not uninteresting:
 
2312
                                    yield result
 
2313
                            advance_path = False
 
2314
                    else:
 
2315
                        result = _process_entry(current_entry, current_path_info)
 
2316
                        if result is not None:
 
2317
                            path_handled = True
 
2318
                            if result is not uninteresting:
 
2319
                                yield result
 
2320
                    if advance_entry and current_entry is not None:
 
2321
                        entry_index += 1
 
2322
                        if entry_index < len(current_block[1]):
 
2323
                            current_entry = current_block[1][entry_index]
 
2324
                        else:
 
2325
                            current_entry = None
 
2326
                    else:
 
2327
                        advance_entry = True # reset the advance flaga
 
2328
                    if advance_path and current_path_info is not None:
 
2329
                        if not path_handled:
 
2330
                            # unversioned in all regards
 
2331
                            if want_unversioned:
 
2332
                                new_executable = bool(
 
2333
                                    stat.S_ISREG(current_path_info[3].st_mode)
 
2334
                                    and stat.S_IEXEC & current_path_info[3].st_mode)
 
2335
                                yield (None,
 
2336
                                    (None, utf8_decode(current_path_info[0])[0]),
 
2337
                                    True,
 
2338
                                    (False, False),
 
2339
                                    (None, None),
 
2340
                                    (None, utf8_decode(current_path_info[1])[0]),
 
2341
                                    (None, current_path_info[2]),
 
2342
                                    (None, new_executable))
 
2343
                            # dont descend into this unversioned path if it is
 
2344
                            # a dir
 
2345
                            if current_path_info[2] in ('directory'):
 
2346
                                del current_dir_info[1][path_index]
 
2347
                                path_index -= 1
 
2348
                        # dont descend the disk iterator into any tree 
 
2349
                        # paths.
 
2350
                        if current_path_info[2] == 'tree-reference':
 
2351
                            del current_dir_info[1][path_index]
 
2352
                            path_index -= 1
 
2353
                        path_index += 1
 
2354
                        if path_index < len(current_dir_info[1]):
 
2355
                            current_path_info = current_dir_info[1][path_index]
 
2356
                            if current_path_info[2] == 'directory':
 
2357
                                if self.target._directory_is_tree_reference(
 
2358
                                    current_path_info[0].decode('utf8')):
 
2359
                                    current_path_info = current_path_info[:2] + \
 
2360
                                        ('tree-reference',) + current_path_info[3:]
 
2361
                        else:
 
2362
                            current_path_info = None
 
2363
                        path_handled = False
 
2364
                    else:
 
2365
                        advance_path = True # reset the advance flagg.
 
2366
                if current_block is not None:
 
2367
                    block_index += 1
 
2368
                    if (block_index < len(state._dirblocks) and
 
2369
                        osutils.is_inside(current_root, state._dirblocks[block_index][0])):
 
2370
                        current_block = state._dirblocks[block_index]
 
2371
                    else:
 
2372
                        current_block = None
 
2373
                if current_dir_info is not None:
 
2374
                    try:
 
2375
                        current_dir_info = dir_iterator.next()
 
2376
                    except StopIteration:
 
2377
                        current_dir_info = None
 
2378
 
 
2379
 
 
2380
    @staticmethod
 
2381
    def is_compatible(source, target):
 
2382
        # the target must be a dirstate working tree
 
2383
        if not isinstance(target, WorkingTree4):
 
2384
            return False
 
2385
        # the source must be a revtreee or dirstate rev tree.
 
2386
        if not isinstance(source,
 
2387
            (revisiontree.RevisionTree, DirStateRevisionTree)):
 
2388
            return False
 
2389
        # the source revid must be in the target dirstate
 
2390
        if not (source._revision_id == NULL_REVISION or
 
2391
            source._revision_id in target.get_parent_ids()):
 
2392
            # TODO: what about ghosts? it may well need to 
 
2393
            # check for them explicitly.
 
2394
            return False
 
2395
        return True
 
2396
 
 
2397
InterTree.register_optimiser(InterDirStateTree)
 
2398
 
 
2399
 
 
2400
class Converter3to4(object):
 
2401
    """Perform an in-place upgrade of format 3 to format 4 trees."""
 
2402
 
 
2403
    def __init__(self):
 
2404
        self.target_format = WorkingTreeFormat4()
 
2405
 
 
2406
    def convert(self, tree):
 
2407
        # lock the control files not the tree, so that we dont get tree
 
2408
        # on-unlock behaviours, and so that noone else diddles with the 
 
2409
        # tree during upgrade.
 
2410
        tree._control_files.lock_write()
 
2411
        try:
 
2412
            tree.read_working_inventory()
 
2413
            self.create_dirstate_data(tree)
 
2414
            self.update_format(tree)
 
2415
            self.remove_xml_files(tree)
 
2416
        finally:
 
2417
            tree._control_files.unlock()
 
2418
 
 
2419
    def create_dirstate_data(self, tree):
 
2420
        """Create the dirstate based data for tree."""
 
2421
        local_path = tree.bzrdir.get_workingtree_transport(None
 
2422
            ).local_abspath('dirstate')
 
2423
        state = dirstate.DirState.from_tree(tree, local_path)
 
2424
        state.save()
 
2425
        state.unlock()
 
2426
 
 
2427
    def remove_xml_files(self, tree):
 
2428
        """Remove the oldformat 3 data."""
 
2429
        transport = tree.bzrdir.get_workingtree_transport(None)
 
2430
        for path in ['basis-inventory-cache', 'inventory', 'last-revision',
 
2431
            'pending-merges', 'stat-cache']:
 
2432
            try:
 
2433
                transport.delete(path)
 
2434
            except errors.NoSuchFile:
 
2435
                # some files are optional - just deal.
 
2436
                pass
 
2437
 
 
2438
    def update_format(self, tree):
 
2439
        """Change the format marker."""
 
2440
        tree._control_files.put_utf8('format',
 
2441
            self.target_format.get_format_string())