1
# Copyright (C) 2005 Canonical Ltd
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.
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.
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
17
"""Tree classes, representing directory at point in time.
21
from collections import deque
22
from cStringIO import StringIO
30
from bzrlib.decorators import needs_read_lock
31
from bzrlib.errors import BzrError, BzrCheckError
32
from bzrlib import errors
33
from bzrlib.inventory import Inventory, InventoryFile
34
from bzrlib.inter import InterObject
35
from bzrlib.osutils import fingerprint_file
36
import bzrlib.revision
37
from bzrlib.trace import mutter, note
41
"""Abstract file tree.
43
There are several subclasses:
45
* `WorkingTree` exists as files on disk editable by the user.
47
* `RevisionTree` is a tree as recorded at some point in the past.
49
Trees contain an `Inventory` object, and also know how to retrieve
50
file texts mentioned in the inventory, either from a working
51
directory or from a store.
53
It is possible for trees to contain files that are not described
54
in their inventory or vice versa; for this use `filenames()`.
56
Trees can be compared, etc, regardless of whether they are working
57
trees or versioned trees.
60
def changes_from(self, other, want_unchanged=False, specific_files=None,
61
extra_trees=None, require_versioned=False, include_root=False,
62
want_unversioned=False):
63
"""Return a TreeDelta of the changes from other to this tree.
65
:param other: A tree to compare with.
66
:param specific_files: An optional list of file paths to restrict the
67
comparison to. When mapping filenames to ids, all matches in all
68
trees (including optional extra_trees) are used, and all children of
69
matched directories are included.
70
:param want_unchanged: An optional boolean requesting the inclusion of
71
unchanged entries in the result.
72
:param extra_trees: An optional list of additional trees to use when
73
mapping the contents of specific_files (paths) to file_ids.
74
:param require_versioned: An optional boolean (defaults to False). When
75
supplied and True all the 'specific_files' must be versioned, or
76
a PathsNotVersionedError will be thrown.
77
:param want_unversioned: Scan for unversioned paths.
79
The comparison will be performed by an InterTree object looked up on
82
# Martin observes that Tree.changes_from returns a TreeDelta and this
83
# may confuse people, because the class name of the returned object is
84
# a synonym of the object referenced in the method name.
85
return InterTree.get(other, self).compare(
86
want_unchanged=want_unchanged,
87
specific_files=specific_files,
88
extra_trees=extra_trees,
89
require_versioned=require_versioned,
90
include_root=include_root,
91
want_unversioned=want_unversioned,
94
def _iter_changes(self, from_tree, include_unchanged=False,
95
specific_files=None, pb=None, extra_trees=None,
96
require_versioned=True, want_unversioned=False):
97
intertree = InterTree.get(from_tree, self)
98
return intertree._iter_changes(include_unchanged, specific_files, pb,
99
extra_trees, require_versioned, want_unversioned=want_unversioned)
102
"""Get a list of the conflicts in the tree.
104
Each conflict is an instance of bzrlib.conflicts.Conflict.
109
"""For trees that can have unversioned files, return all such paths."""
112
def get_parent_ids(self):
113
"""Get the parent ids for this tree.
115
:return: a list of parent ids. [] is returned to indicate
116
a tree with no parents.
117
:raises: BzrError if the parents are not known.
119
raise NotImplementedError(self.get_parent_ids)
121
def has_filename(self, filename):
122
"""True if the tree has given filename."""
123
raise NotImplementedError()
125
def has_id(self, file_id):
126
file_id = osutils.safe_file_id(file_id)
127
return self.inventory.has_id(file_id)
129
__contains__ = has_id
131
def has_or_had_id(self, file_id):
132
file_id = osutils.safe_file_id(file_id)
133
if file_id == self.inventory.root.file_id:
135
return self.inventory.has_id(file_id)
138
return iter(self.inventory)
140
def id2path(self, file_id):
141
file_id = osutils.safe_file_id(file_id)
142
return self.inventory.id2path(file_id)
144
def is_control_filename(self, filename):
145
"""True if filename is the name of a control file in this tree.
147
:param filename: A filename within the tree. This is a relative path
148
from the root of this tree.
150
This is true IF and ONLY IF the filename is part of the meta data
151
that bzr controls in this tree. I.E. a random .bzr directory placed
152
on disk will not be a control file for this tree.
154
return self.bzrdir.is_control_filename(filename)
157
def iter_entries_by_dir(self, specific_file_ids=None):
158
"""Walk the tree in 'by_dir' order.
160
This will yield each entry in the tree as a (path, entry) tuple. The
161
order that they are yielded is: the contents of a directory are
162
preceeded by the parent of a directory, and all the contents of a
163
directory are grouped together.
165
return self.inventory.iter_entries_by_dir(
166
specific_file_ids=specific_file_ids)
168
def iter_reference_entries(self):
169
for path, entry in self.iter_entries_by_dir():
170
if entry.kind == 'tree-reference':
173
def kind(self, file_id):
174
raise NotImplementedError("Tree subclass %s must implement kind"
175
% self.__class__.__name__)
177
def get_reference_revision(self, entry, path=None):
178
raise NotImplementedError("Tree subclass %s must implement "
179
"get_reference_revision"
180
% self.__class__.__name__)
182
def _comparison_data(self, entry, path):
183
"""Return a tuple of kind, executable, stat_value for a file.
185
entry may be None if there is no inventory entry for the file, but
186
path must always be supplied.
188
kind is None if there is no file present (even if an inventory id is
189
present). executable is False for non-file entries.
191
raise NotImplementedError(self._comparison_data)
193
def _file_size(self, entry, stat_value):
194
raise NotImplementedError(self._file_size)
196
def _get_inventory(self):
197
return self._inventory
199
def get_file(self, file_id):
200
"""Return a file object for the file file_id in the tree."""
201
raise NotImplementedError(self.get_file)
203
def get_file_mtime(self, file_id, path=None):
204
"""Return the modification time for a file.
206
:param file_id: The handle for this file.
207
:param path: The path that this file can be found at.
208
These must point to the same object.
210
raise NotImplementedError(self.get_file_mtime)
212
def get_file_by_path(self, path):
213
return self.get_file(self._inventory.path2id(path))
215
def get_symlink_target(self, file_id):
216
"""Get the target for a given file_id.
218
It is assumed that the caller already knows that file_id is referencing
220
:param file_id: Handle for the symlink entry.
221
:return: The path the symlink points to.
223
raise NotImplementedError(self.get_symlink_target)
225
def annotate_iter(self, file_id):
226
"""Return an iterator of revision_id, line tuples
228
For working trees (and mutable trees in general), the special
229
revision_id 'current:' will be used for lines that are new in this
230
tree, e.g. uncommitted changes.
231
:param file_id: The file to produce an annotated version from
233
raise NotImplementedError(self.annotate_iter)
235
inventory = property(_get_inventory,
236
doc="Inventory of this Tree")
238
def _check_retrieved(self, ie, f):
241
fp = fingerprint_file(f)
244
if ie.text_size is not None:
245
if ie.text_size != fp['size']:
246
raise BzrError("mismatched size for file %r in %r" % (ie.file_id, self._store),
247
["inventory expects %d bytes" % ie.text_size,
248
"file is actually %d bytes" % fp['size'],
249
"store is probably damaged/corrupt"])
251
if ie.text_sha1 != fp['sha1']:
252
raise BzrError("wrong SHA-1 for file %r in %r" % (ie.file_id, self._store),
253
["inventory expects %s" % ie.text_sha1,
254
"file is actually %s" % fp['sha1'],
255
"store is probably damaged/corrupt"])
257
def path2id(self, path):
258
"""Return the id for path in this tree."""
259
return self._inventory.path2id(path)
261
def paths2ids(self, paths, trees=[], require_versioned=True):
262
"""Return all the ids that can be reached by walking from paths.
264
Each path is looked up in each this tree and any extras provided in
265
trees, and this is repeated recursively: the children in an extra tree
266
of a directory that has been renamed under a provided path in this tree
267
are all returned, even if none exist until a provided path in this
268
tree, and vice versa.
270
:param paths: An iterable of paths to start converting to ids from.
271
Alternatively, if paths is None, no ids should be calculated and None
272
will be returned. This is offered to make calling the api unconditional
273
for code that *might* take a list of files.
274
:param trees: Additional trees to consider.
275
:param require_versioned: If False, do not raise NotVersionedError if
276
an element of paths is not versioned in this tree and all of trees.
278
return find_ids_across_trees(paths, [self] + list(trees), require_versioned)
280
def print_file(self, file_id):
281
"""Print file with id `file_id` to stdout."""
282
file_id = osutils.safe_file_id(file_id)
284
sys.stdout.write(self.get_file_text(file_id))
289
def revision_tree(self, revision_id):
290
"""Obtain a revision tree for the revision revision_id.
292
The intention of this method is to allow access to possibly cached
293
tree data. Implementors of this method should raise NoSuchRevision if
294
the tree is not locally available, even if they could obtain the
295
tree via a repository or some other means. Callers are responsible
296
for finding the ultimate source for a revision tree.
298
:param revision_id: The revision_id of the requested tree.
300
:raises: NoSuchRevision if the tree cannot be obtained.
302
raise errors.NoSuchRevisionInTree(self, revision_id)
305
"""What files are present in this tree and unknown.
307
:return: an iterator over the unknown files.
314
def filter_unversioned_files(self, paths):
315
"""Filter out paths that are versioned.
317
:return: set of paths.
319
# NB: we specifically *don't* call self.has_filename, because for
320
# WorkingTrees that can indicate files that exist on disk but that
322
pred = self.inventory.has_filename
323
return set((p for p in paths if not pred(p)))
325
def walkdirs(self, prefix=""):
326
"""Walk the contents of this tree from path down.
328
This yields all the data about the contents of a directory at a time.
329
After each directory has been yielded, if the caller has mutated the
330
list to exclude some directories, they are then not descended into.
332
The data yielded is of the form:
333
((directory-relpath, directory-path-from-root, directory-fileid),
334
[(relpath, basename, kind, lstat, path_from_tree_root, file_id,
335
versioned_kind), ...]),
336
- directory-relpath is the containing dirs relpath from prefix
337
- directory-path-from-root is the containing dirs path from /
338
- directory-fileid is the id of the directory if it is versioned.
339
- relpath is the relative path within the subtree being walked.
340
- basename is the basename
341
- kind is the kind of the file now. If unknonwn then the file is not
342
present within the tree - but it may be recorded as versioned. See
344
- lstat is the stat data *if* the file was statted.
345
- path_from_tree_root is the path from the root of the tree.
346
- file_id is the file_id is the entry is versioned.
347
- versioned_kind is the kind of the file as last recorded in the
348
versioning system. If 'unknown' the file is not versioned.
349
One of 'kind' and 'versioned_kind' must not be 'unknown'.
351
:param prefix: Start walking from prefix within the tree rather than
352
at the root. This allows one to walk a subtree but get paths that are
353
relative to a tree rooted higher up.
354
:return: an iterator over the directory data.
356
raise NotImplementedError(self.walkdirs)
359
class EmptyTree(Tree):
362
self._inventory = Inventory(root_id=None)
363
symbol_versioning.warn('EmptyTree is deprecated as of bzr 0.9 please'
364
' use repository.revision_tree instead.',
365
DeprecationWarning, stacklevel=2)
367
def get_parent_ids(self):
370
def get_symlink_target(self, file_id):
373
def has_filename(self, filename):
376
def kind(self, file_id):
377
file_id = osutils.safe_file_id(file_id)
378
assert self._inventory[file_id].kind == "directory"
381
def list_files(self, include_root=False):
384
def __contains__(self, file_id):
385
file_id = osutils.safe_file_id(file_id)
386
return (file_id in self._inventory)
388
def get_file_sha1(self, file_id, path=None, stat_value=None):
392
######################################################################
395
# TODO: Merge these two functions into a single one that can operate
396
# on either a whole tree or a set of files.
398
# TODO: Return the diff in order by filename, not by category or in
399
# random order. Can probably be done by lock-stepping through the
400
# filenames from both trees.
403
def file_status(filename, old_tree, new_tree):
404
"""Return single-letter status, old and new names for a file.
406
The complexity here is in deciding how to represent renames;
407
many complex cases are possible.
409
old_inv = old_tree.inventory
410
new_inv = new_tree.inventory
411
new_id = new_inv.path2id(filename)
412
old_id = old_inv.path2id(filename)
414
if not new_id and not old_id:
415
# easy: doesn't exist in either; not versioned at all
416
if new_tree.is_ignored(filename):
417
return 'I', None, None
419
return '?', None, None
421
# There is now a file of this name, great.
424
# There is no longer a file of this name, but we can describe
425
# what happened to the file that used to have
426
# this name. There are two possibilities: either it was
427
# deleted entirely, or renamed.
429
if new_inv.has_id(old_id):
430
return 'X', old_inv.id2path(old_id), new_inv.id2path(old_id)
432
return 'D', old_inv.id2path(old_id), None
434
# if the file_id is new in this revision, it is added
435
if new_id and not old_inv.has_id(new_id):
438
# if there used to be a file of this name, but that ID has now
439
# disappeared, it is deleted
440
if old_id and not new_inv.has_id(old_id):
447
def find_renames(old_inv, new_inv):
448
for file_id in old_inv:
449
if file_id not in new_inv:
451
old_name = old_inv.id2path(file_id)
452
new_name = new_inv.id2path(file_id)
453
if old_name != new_name:
454
yield (old_name, new_name)
457
def find_ids_across_trees(filenames, trees, require_versioned=True):
458
"""Find the ids corresponding to specified filenames.
460
All matches in all trees will be used, and all children of matched
461
directories will be used.
463
:param filenames: The filenames to find file_ids for (if None, returns
465
:param trees: The trees to find file_ids within
466
:param require_versioned: if true, all specified filenames must occur in
468
:return: a set of file ids for the specified filenames and their children.
472
specified_path_ids = _find_ids_across_trees(filenames, trees,
474
return _find_children_across_trees(specified_path_ids, trees)
477
def _find_ids_across_trees(filenames, trees, require_versioned):
478
"""Find the ids corresponding to specified filenames.
480
All matches in all trees will be used, but subdirectories are not scanned.
482
:param filenames: The filenames to find file_ids for
483
:param trees: The trees to find file_ids within
484
:param require_versioned: if true, all specified filenames must occur in
486
:return: a set of (path, file ids) for the specified filenames
489
interesting_ids = set()
490
for tree_path in filenames:
493
file_id = tree.path2id(tree_path)
494
if file_id is not None:
495
interesting_ids.add(file_id)
498
not_versioned.append(tree_path)
499
if len(not_versioned) > 0 and require_versioned:
500
raise errors.PathsNotVersionedError(not_versioned)
501
return interesting_ids
504
def _find_children_across_trees(specified_ids, trees):
505
"""Return a set including specified ids and their children
507
All matches in all trees will be used.
509
:param trees: The trees to find file_ids within
510
:return: a set containing all specified ids and their children
512
interesting_ids = set(specified_ids)
513
pending = interesting_ids
514
# now handle children of interesting ids
515
# we loop so that we handle all children of each id in both trees
516
while len(pending) > 0:
518
for file_id in pending:
520
if not tree.has_id(file_id):
522
entry = tree.inventory[file_id]
523
for child in getattr(entry, 'children', {}).itervalues():
524
if child.file_id not in interesting_ids:
525
new_pending.add(child.file_id)
526
interesting_ids.update(new_pending)
527
pending = new_pending
528
return interesting_ids
531
class InterTree(InterObject):
532
"""This class represents operations taking place between two Trees.
534
Its instances have methods like 'compare' and contain references to the
535
source and target trees these operations are to be carried out on.
537
clients of bzrlib should not need to use InterTree directly, rather they
538
should use the convenience methods on Tree such as 'Tree.compare()' which
539
will pass through to InterTree as appropriate.
545
def compare(self, want_unchanged=False, specific_files=None,
546
extra_trees=None, require_versioned=False, include_root=False,
547
want_unversioned=False):
548
"""Return the changes from source to target.
550
:return: A TreeDelta.
551
:param specific_files: An optional list of file paths to restrict the
552
comparison to. When mapping filenames to ids, all matches in all
553
trees (including optional extra_trees) are used, and all children of
554
matched directories are included.
555
:param want_unchanged: An optional boolean requesting the inclusion of
556
unchanged entries in the result.
557
:param extra_trees: An optional list of additional trees to use when
558
mapping the contents of specific_files (paths) to file_ids.
559
:param require_versioned: An optional boolean (defaults to False). When
560
supplied and True all the 'specific_files' must be versioned, or
561
a PathsNotVersionedError will be thrown.
562
:param want_unversioned: Scan for unversioned paths.
564
# NB: show_status depends on being able to pass in non-versioned files
565
# and report them as unknown
566
trees = (self.source,)
567
if extra_trees is not None:
568
trees = trees + tuple(extra_trees)
569
# target is usually the newer tree:
570
specific_file_ids = self.target.paths2ids(specific_files, trees,
571
require_versioned=require_versioned)
572
if specific_files and not specific_file_ids:
573
# All files are unversioned, so just return an empty delta
574
# _compare_trees would think we want a complete delta
575
result = delta.TreeDelta()
576
fake_entry = InventoryFile('unused', 'unused', 'unused')
577
result.unversioned = [(path, None,
578
self.target._comparison_data(fake_entry, path)[0]) for path in
581
return delta._compare_trees(self.source, self.target, want_unchanged,
582
specific_files, include_root, extra_trees=extra_trees,
583
want_unversioned=want_unversioned)
585
def _iter_changes(self, include_unchanged=False,
586
specific_files=None, pb=None, extra_trees=[],
587
require_versioned=True, want_unversioned=False):
588
"""Generate an iterator of changes between trees.
591
(file_id, path, changed_content, versioned, parent, name, kind,
594
Path is relative to the target tree. changed_content is True if the
595
file's content has changed. This includes changes to its kind, and to
598
versioned, parent, name, kind, executable are tuples of (from, to).
599
If a file is missing in a tree, its kind is None.
601
Iteration is done in parent-to-child order, relative to the target
604
There is no guarantee that all paths are in sorted order: the
605
requirement to expand the search due to renames may result in children
606
that should be found early being found late in the search, after
607
lexically later results have been returned.
608
:param require_versioned: Raise errors.PathsNotVersionedError if a
609
path in the specific_files list is not versioned in one of
610
source, target or extra_trees.
611
:param want_unversioned: Should unversioned files be returned in the
612
output. An unversioned file is defined as one with (False, False)
613
for the versioned pair.
615
# this must return a sequence rather than a list so that it can hold a
616
# read-lock for the whole time.
618
# TODO: this really only needs to lock the trees not the branches, so
619
# could do with lock_tree_read() -- mbp 20070227
621
## self.source.lock_read()
622
## self.target.lock_read()
624
lookup_trees = [self.source]
626
lookup_trees.extend(extra_trees)
627
specific_file_ids = self.target.paths2ids(specific_files,
628
lookup_trees, require_versioned=require_versioned)
630
all_unversioned = sorted([(p.split('/'), p) for p in self.target.extras()
631
if not specific_files or
632
osutils.is_inside_any(specific_files, p)])
633
all_unversioned = deque(all_unversioned)
635
all_unversioned = deque()
637
from_entries_by_dir = list(self.source.inventory.iter_entries_by_dir(
638
specific_file_ids=specific_file_ids))
639
from_data = dict((e.file_id, (p, e)) for p, e in from_entries_by_dir)
640
to_entries_by_dir = list(self.target.inventory.iter_entries_by_dir(
641
specific_file_ids=specific_file_ids))
642
num_entries = len(from_entries_by_dir) + len(to_entries_by_dir)
644
# the unversioned path lookup only occurs on real trees - where there
645
# can be extras. So the fake_entry is solely used to look up
646
# executable it values when execute is not supported.
647
fake_entry = InventoryFile('unused', 'unused', 'unused')
648
for to_path, to_entry in to_entries_by_dir:
649
while all_unversioned and all_unversioned[0][0] < to_path.split('/'):
650
unversioned_path = all_unversioned.popleft()
651
to_kind, to_executable, to_stat = \
652
self.target._comparison_data(fake_entry, unversioned_path[1])
653
yield (None, unversioned_path[1], True, (False, False),
655
(None, unversioned_path[0][-1]),
657
(None, to_executable))
658
file_id = to_entry.file_id
659
to_paths[file_id] = to_path
661
changed_content = False
662
from_path, from_entry = from_data.get(file_id, (None, None))
663
from_versioned = (from_entry is not None)
664
if from_entry is not None:
665
from_versioned = True
666
from_name = from_entry.name
667
from_parent = from_entry.parent_id
668
from_kind, from_executable, from_stat = \
669
self.source._comparison_data(from_entry, from_path)
672
from_versioned = False
676
from_executable = None
677
versioned = (from_versioned, True)
678
to_kind, to_executable, to_stat = \
679
self.target._comparison_data(to_entry, to_path)
680
kind = (from_kind, to_kind)
681
if kind[0] != kind[1]:
682
changed_content = True
683
elif from_kind == 'file':
684
from_size = self.source._file_size(from_entry, from_stat)
685
to_size = self.target._file_size(to_entry, to_stat)
686
if from_size != to_size:
687
changed_content = True
688
elif (self.source.get_file_sha1(file_id, from_path, from_stat) !=
689
self.target.get_file_sha1(file_id, to_path, to_stat)):
690
changed_content = True
691
elif from_kind == 'symlink':
692
if (self.source.get_symlink_target(file_id) !=
693
self.target.get_symlink_target(file_id)):
694
changed_content = True
695
elif from_kind == 'tree-reference':
696
if (self.source.get_reference_revision(from_entry, from_path)
697
!= self.target.get_reference_revision(to_entry, to_path)):
698
changed_content = True
699
parent = (from_parent, to_entry.parent_id)
700
name = (from_name, to_entry.name)
701
executable = (from_executable, to_executable)
703
pb.update('comparing files', entry_count, num_entries)
704
if (changed_content is not False or versioned[0] != versioned[1]
705
or parent[0] != parent[1] or name[0] != name[1] or
706
executable[0] != executable[1] or include_unchanged):
707
yield (file_id, to_path, changed_content, versioned, parent,
708
name, kind, executable)
709
while all_unversioned:
710
# yield any trailing unversioned paths
711
unversioned_path = all_unversioned.popleft()
712
to_kind, to_executable, to_stat = \
713
self.target._comparison_data(fake_entry, unversioned_path[1])
714
yield (None, unversioned_path[1], True, (False, False),
716
(None, unversioned_path[0][-1]),
718
(None, to_executable))
720
def get_to_path(from_entry):
721
if from_entry.parent_id is None:
724
if from_entry.parent_id not in to_paths:
725
get_to_path(self.source.inventory[from_entry.parent_id])
726
to_path = osutils.pathjoin(to_paths[from_entry.parent_id],
728
to_paths[from_entry.file_id] = to_path
731
for path, from_entry in from_entries_by_dir:
732
file_id = from_entry.file_id
733
if file_id in to_paths:
735
to_path = get_to_path(from_entry)
738
pb.update('comparing files', entry_count, num_entries)
739
versioned = (True, False)
740
parent = (from_entry.parent_id, None)
741
name = (from_entry.name, None)
742
from_kind, from_executable, stat_value = \
743
self.source._comparison_data(from_entry, path)
744
kind = (from_kind, None)
745
executable = (from_executable, None)
746
changed_content = True
747
# the parent's path is necessarily known at this point.
748
yield(file_id, to_path, changed_content, versioned, parent,
749
name, kind, executable)
752
# This was deprecated before 0.12, but did not have an official warning
753
@symbol_versioning.deprecated_function(symbol_versioning.zero_twelve)
754
def RevisionTree(*args, **kwargs):
755
"""RevisionTree has moved to bzrlib.revisiontree.RevisionTree()
757
Accessing it as bzrlib.tree.RevisionTree has been deprecated as of
760
from bzrlib.revisiontree import RevisionTree as _RevisionTree
761
return _RevisionTree(*args, **kwargs)
added
removed
bzrlib/tree.py
