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)
137
def is_ignored(self, filename):
138
"""Check whether the filename is ignored by this tree.
140
:param filename: The relative filename within the tree.
141
:return: True if the filename is ignored.
146
return iter(self.inventory)
148
def id2path(self, file_id):
149
"""Return the path for a file id.
153
file_id = osutils.safe_file_id(file_id)
154
return self.inventory.id2path(file_id)
156
def is_control_filename(self, filename):
157
"""True if filename is the name of a control file in this tree.
159
:param filename: A filename within the tree. This is a relative path
160
from the root of this tree.
162
This is true IF and ONLY IF the filename is part of the meta data
163
that bzr controls in this tree. I.E. a random .bzr directory placed
164
on disk will not be a control file for this tree.
166
return self.bzrdir.is_control_filename(filename)
169
def iter_entries_by_dir(self, specific_file_ids=None):
170
"""Walk the tree in 'by_dir' order.
172
This will yield each entry in the tree as a (path, entry) tuple. The
173
order that they are yielded is: the contents of a directory are
174
preceeded by the parent of a directory, and all the contents of a
175
directory are grouped together.
177
return self.inventory.iter_entries_by_dir(
178
specific_file_ids=specific_file_ids)
180
def iter_references(self):
181
for path, entry in self.iter_entries_by_dir():
182
if entry.kind == 'tree-reference':
183
yield path, entry.file_id
185
def kind(self, file_id):
186
raise NotImplementedError("Tree subclass %s must implement kind"
187
% self.__class__.__name__)
189
def get_reference_revision(self, file_id, path=None):
190
raise NotImplementedError("Tree subclass %s must implement "
191
"get_reference_revision"
192
% self.__class__.__name__)
194
def _comparison_data(self, entry, path):
195
"""Return a tuple of kind, executable, stat_value for a file.
197
entry may be None if there is no inventory entry for the file, but
198
path must always be supplied.
200
kind is None if there is no file present (even if an inventory id is
201
present). executable is False for non-file entries.
203
raise NotImplementedError(self._comparison_data)
205
def _file_size(self, entry, stat_value):
206
raise NotImplementedError(self._file_size)
208
def _get_inventory(self):
209
return self._inventory
211
def get_file(self, file_id):
212
"""Return a file object for the file file_id in the tree."""
213
raise NotImplementedError(self.get_file)
215
def get_file_mtime(self, file_id, path=None):
216
"""Return the modification time for a file.
218
:param file_id: The handle for this file.
219
:param path: The path that this file can be found at.
220
These must point to the same object.
222
raise NotImplementedError(self.get_file_mtime)
224
def get_file_by_path(self, path):
225
return self.get_file(self._inventory.path2id(path))
227
def get_symlink_target(self, file_id):
228
"""Get the target for a given file_id.
230
It is assumed that the caller already knows that file_id is referencing
232
:param file_id: Handle for the symlink entry.
233
:return: The path the symlink points to.
235
raise NotImplementedError(self.get_symlink_target)
237
def annotate_iter(self, file_id):
238
"""Return an iterator of revision_id, line tuples
240
For working trees (and mutable trees in general), the special
241
revision_id 'current:' will be used for lines that are new in this
242
tree, e.g. uncommitted changes.
243
:param file_id: The file to produce an annotated version from
245
raise NotImplementedError(self.annotate_iter)
247
inventory = property(_get_inventory,
248
doc="Inventory of this Tree")
250
def _check_retrieved(self, ie, f):
253
fp = fingerprint_file(f)
256
if ie.text_size is not None:
257
if ie.text_size != fp['size']:
258
raise BzrError("mismatched size for file %r in %r" % (ie.file_id, self._store),
259
["inventory expects %d bytes" % ie.text_size,
260
"file is actually %d bytes" % fp['size'],
261
"store is probably damaged/corrupt"])
263
if ie.text_sha1 != fp['sha1']:
264
raise BzrError("wrong SHA-1 for file %r in %r" % (ie.file_id, self._store),
265
["inventory expects %s" % ie.text_sha1,
266
"file is actually %s" % fp['sha1'],
267
"store is probably damaged/corrupt"])
270
def path2id(self, path):
271
"""Return the id for path in this tree."""
272
return self._inventory.path2id(path)
274
def paths2ids(self, paths, trees=[], require_versioned=True):
275
"""Return all the ids that can be reached by walking from paths.
277
Each path is looked up in each this tree and any extras provided in
278
trees, and this is repeated recursively: the children in an extra tree
279
of a directory that has been renamed under a provided path in this tree
280
are all returned, even if none exist until a provided path in this
281
tree, and vice versa.
283
:param paths: An iterable of paths to start converting to ids from.
284
Alternatively, if paths is None, no ids should be calculated and None
285
will be returned. This is offered to make calling the api unconditional
286
for code that *might* take a list of files.
287
:param trees: Additional trees to consider.
288
:param require_versioned: If False, do not raise NotVersionedError if
289
an element of paths is not versioned in this tree and all of trees.
291
return find_ids_across_trees(paths, [self] + list(trees), require_versioned)
293
def print_file(self, file_id):
294
"""Print file with id `file_id` to stdout."""
295
file_id = osutils.safe_file_id(file_id)
297
sys.stdout.write(self.get_file_text(file_id))
302
def revision_tree(self, revision_id):
303
"""Obtain a revision tree for the revision revision_id.
305
The intention of this method is to allow access to possibly cached
306
tree data. Implementors of this method should raise NoSuchRevision if
307
the tree is not locally available, even if they could obtain the
308
tree via a repository or some other means. Callers are responsible
309
for finding the ultimate source for a revision tree.
311
:param revision_id: The revision_id of the requested tree.
313
:raises: NoSuchRevision if the tree cannot be obtained.
315
raise errors.NoSuchRevisionInTree(self, revision_id)
318
"""What files are present in this tree and unknown.
320
:return: an iterator over the unknown files.
327
def filter_unversioned_files(self, paths):
328
"""Filter out paths that are versioned.
330
:return: set of paths.
332
# NB: we specifically *don't* call self.has_filename, because for
333
# WorkingTrees that can indicate files that exist on disk but that
335
pred = self.inventory.has_filename
336
return set((p for p in paths if not pred(p)))
338
def walkdirs(self, prefix=""):
339
"""Walk the contents of this tree from path down.
341
This yields all the data about the contents of a directory at a time.
342
After each directory has been yielded, if the caller has mutated the
343
list to exclude some directories, they are then not descended into.
345
The data yielded is of the form:
346
((directory-relpath, directory-path-from-root, directory-fileid),
347
[(relpath, basename, kind, lstat, path_from_tree_root, file_id,
348
versioned_kind), ...]),
349
- directory-relpath is the containing dirs relpath from prefix
350
- directory-path-from-root is the containing dirs path from /
351
- directory-fileid is the id of the directory if it is versioned.
352
- relpath is the relative path within the subtree being walked.
353
- basename is the basename
354
- kind is the kind of the file now. If unknonwn then the file is not
355
present within the tree - but it may be recorded as versioned. See
357
- lstat is the stat data *if* the file was statted.
358
- path_from_tree_root is the path from the root of the tree.
359
- file_id is the file_id is the entry is versioned.
360
- versioned_kind is the kind of the file as last recorded in the
361
versioning system. If 'unknown' the file is not versioned.
362
One of 'kind' and 'versioned_kind' must not be 'unknown'.
364
:param prefix: Start walking from prefix within the tree rather than
365
at the root. This allows one to walk a subtree but get paths that are
366
relative to a tree rooted higher up.
367
:return: an iterator over the directory data.
369
raise NotImplementedError(self.walkdirs)
372
class EmptyTree(Tree):
375
self._inventory = Inventory(root_id=None)
376
symbol_versioning.warn('EmptyTree is deprecated as of bzr 0.9 please'
377
' use repository.revision_tree instead.',
378
DeprecationWarning, stacklevel=2)
380
def get_parent_ids(self):
383
def get_symlink_target(self, file_id):
386
def has_filename(self, filename):
389
def kind(self, file_id):
390
file_id = osutils.safe_file_id(file_id)
391
assert self._inventory[file_id].kind == "directory"
394
def list_files(self, include_root=False):
397
def __contains__(self, file_id):
398
file_id = osutils.safe_file_id(file_id)
399
return (file_id in self._inventory)
401
def get_file_sha1(self, file_id, path=None, stat_value=None):
405
######################################################################
408
# TODO: Merge these two functions into a single one that can operate
409
# on either a whole tree or a set of files.
411
# TODO: Return the diff in order by filename, not by category or in
412
# random order. Can probably be done by lock-stepping through the
413
# filenames from both trees.
416
def file_status(filename, old_tree, new_tree):
417
"""Return single-letter status, old and new names for a file.
419
The complexity here is in deciding how to represent renames;
420
many complex cases are possible.
422
old_inv = old_tree.inventory
423
new_inv = new_tree.inventory
424
new_id = new_inv.path2id(filename)
425
old_id = old_inv.path2id(filename)
427
if not new_id and not old_id:
428
# easy: doesn't exist in either; not versioned at all
429
if new_tree.is_ignored(filename):
430
return 'I', None, None
432
return '?', None, None
434
# There is now a file of this name, great.
437
# There is no longer a file of this name, but we can describe
438
# what happened to the file that used to have
439
# this name. There are two possibilities: either it was
440
# deleted entirely, or renamed.
442
if new_inv.has_id(old_id):
443
return 'X', old_inv.id2path(old_id), new_inv.id2path(old_id)
445
return 'D', old_inv.id2path(old_id), None
447
# if the file_id is new in this revision, it is added
448
if new_id and not old_inv.has_id(new_id):
451
# if there used to be a file of this name, but that ID has now
452
# disappeared, it is deleted
453
if old_id and not new_inv.has_id(old_id):
460
def find_renames(old_inv, new_inv):
461
for file_id in old_inv:
462
if file_id not in new_inv:
464
old_name = old_inv.id2path(file_id)
465
new_name = new_inv.id2path(file_id)
466
if old_name != new_name:
467
yield (old_name, new_name)
470
def find_ids_across_trees(filenames, trees, require_versioned=True):
471
"""Find the ids corresponding to specified filenames.
473
All matches in all trees will be used, and all children of matched
474
directories will be used.
476
:param filenames: The filenames to find file_ids for (if None, returns
478
:param trees: The trees to find file_ids within
479
:param require_versioned: if true, all specified filenames must occur in
481
:return: a set of file ids for the specified filenames and their children.
485
specified_path_ids = _find_ids_across_trees(filenames, trees,
487
return _find_children_across_trees(specified_path_ids, trees)
490
def _find_ids_across_trees(filenames, trees, require_versioned):
491
"""Find the ids corresponding to specified filenames.
493
All matches in all trees will be used, but subdirectories are not scanned.
495
:param filenames: The filenames to find file_ids for
496
:param trees: The trees to find file_ids within
497
:param require_versioned: if true, all specified filenames must occur in
499
:return: a set of (path, file ids) for the specified filenames
502
interesting_ids = set()
503
for tree_path in filenames:
506
file_id = tree.path2id(tree_path)
507
if file_id is not None:
508
interesting_ids.add(file_id)
511
not_versioned.append(tree_path)
512
if len(not_versioned) > 0 and require_versioned:
513
raise errors.PathsNotVersionedError(not_versioned)
514
return interesting_ids
517
def _find_children_across_trees(specified_ids, trees):
518
"""Return a set including specified ids and their children
520
All matches in all trees will be used.
522
:param trees: The trees to find file_ids within
523
:return: a set containing all specified ids and their children
525
interesting_ids = set(specified_ids)
526
pending = interesting_ids
527
# now handle children of interesting ids
528
# we loop so that we handle all children of each id in both trees
529
while len(pending) > 0:
531
for file_id in pending:
533
if not tree.has_id(file_id):
535
entry = tree.inventory[file_id]
536
for child in getattr(entry, 'children', {}).itervalues():
537
if child.file_id not in interesting_ids:
538
new_pending.add(child.file_id)
539
interesting_ids.update(new_pending)
540
pending = new_pending
541
return interesting_ids
544
class InterTree(InterObject):
545
"""This class represents operations taking place between two Trees.
547
Its instances have methods like 'compare' and contain references to the
548
source and target trees these operations are to be carried out on.
550
clients of bzrlib should not need to use InterTree directly, rather they
551
should use the convenience methods on Tree such as 'Tree.compare()' which
552
will pass through to InterTree as appropriate.
558
def compare(self, want_unchanged=False, specific_files=None,
559
extra_trees=None, require_versioned=False, include_root=False,
560
want_unversioned=False):
561
"""Return the changes from source to target.
563
:return: A TreeDelta.
564
:param specific_files: An optional list of file paths to restrict the
565
comparison to. When mapping filenames to ids, all matches in all
566
trees (including optional extra_trees) are used, and all children of
567
matched directories are included.
568
:param want_unchanged: An optional boolean requesting the inclusion of
569
unchanged entries in the result.
570
:param extra_trees: An optional list of additional trees to use when
571
mapping the contents of specific_files (paths) to file_ids.
572
:param require_versioned: An optional boolean (defaults to False). When
573
supplied and True all the 'specific_files' must be versioned, or
574
a PathsNotVersionedError will be thrown.
575
:param want_unversioned: Scan for unversioned paths.
577
# NB: show_status depends on being able to pass in non-versioned files
578
# and report them as unknown
579
trees = (self.source,)
580
if extra_trees is not None:
581
trees = trees + tuple(extra_trees)
582
# target is usually the newer tree:
583
specific_file_ids = self.target.paths2ids(specific_files, trees,
584
require_versioned=require_versioned)
585
if specific_files and not specific_file_ids:
586
# All files are unversioned, so just return an empty delta
587
# _compare_trees would think we want a complete delta
588
result = delta.TreeDelta()
589
fake_entry = InventoryFile('unused', 'unused', 'unused')
590
result.unversioned = [(path, None,
591
self.target._comparison_data(fake_entry, path)[0]) for path in
594
return delta._compare_trees(self.source, self.target, want_unchanged,
595
specific_files, include_root, extra_trees=extra_trees,
596
want_unversioned=want_unversioned)
598
def _iter_changes(self, include_unchanged=False,
599
specific_files=None, pb=None, extra_trees=[],
600
require_versioned=True, want_unversioned=False):
601
"""Generate an iterator of changes between trees.
604
(file_id, (path_in_source, path_in_target),
605
changed_content, versioned, parent, name, kind,
608
Changed_content is True if the file's content has changed. This
609
includes changes to its kind, and to a symlink's target.
611
versioned, parent, name, kind, executable are tuples of (from, to).
612
If a file is missing in a tree, its kind is None.
614
Iteration is done in parent-to-child order, relative to the target
617
There is no guarantee that all paths are in sorted order: the
618
requirement to expand the search due to renames may result in children
619
that should be found early being found late in the search, after
620
lexically later results have been returned.
621
:param require_versioned: Raise errors.PathsNotVersionedError if a
622
path in the specific_files list is not versioned in one of
623
source, target or extra_trees.
624
:param want_unversioned: Should unversioned files be returned in the
625
output. An unversioned file is defined as one with (False, False)
626
for the versioned pair.
629
lookup_trees = [self.source]
631
lookup_trees.extend(extra_trees)
632
specific_file_ids = self.target.paths2ids(specific_files,
633
lookup_trees, require_versioned=require_versioned)
635
all_unversioned = sorted([(p.split('/'), p) for p in self.target.extras()
636
if not specific_files or
637
osutils.is_inside_any(specific_files, p)])
638
all_unversioned = deque(all_unversioned)
640
all_unversioned = deque()
642
from_entries_by_dir = list(self.source.inventory.iter_entries_by_dir(
643
specific_file_ids=specific_file_ids))
644
from_data = dict((e.file_id, (p, e)) for p, e in from_entries_by_dir)
645
to_entries_by_dir = list(self.target.inventory.iter_entries_by_dir(
646
specific_file_ids=specific_file_ids))
647
num_entries = len(from_entries_by_dir) + len(to_entries_by_dir)
649
# the unversioned path lookup only occurs on real trees - where there
650
# can be extras. So the fake_entry is solely used to look up
651
# executable it values when execute is not supported.
652
fake_entry = InventoryFile('unused', 'unused', 'unused')
653
for to_path, to_entry in to_entries_by_dir:
654
while all_unversioned and all_unversioned[0][0] < to_path.split('/'):
655
unversioned_path = all_unversioned.popleft()
656
to_kind, to_executable, to_stat = \
657
self.target._comparison_data(fake_entry, unversioned_path[1])
658
yield (None, (None, unversioned_path[1]), True, (False, False),
660
(None, unversioned_path[0][-1]),
662
(None, to_executable))
663
file_id = to_entry.file_id
664
to_paths[file_id] = to_path
666
changed_content = False
667
from_path, from_entry = from_data.get(file_id, (None, None))
668
from_versioned = (from_entry is not None)
669
if from_entry is not None:
670
from_versioned = True
671
from_name = from_entry.name
672
from_parent = from_entry.parent_id
673
from_kind, from_executable, from_stat = \
674
self.source._comparison_data(from_entry, from_path)
677
from_versioned = False
681
from_executable = None
682
versioned = (from_versioned, True)
683
to_kind, to_executable, to_stat = \
684
self.target._comparison_data(to_entry, to_path)
685
kind = (from_kind, to_kind)
686
if kind[0] != kind[1]:
687
changed_content = True
688
elif from_kind == 'file':
689
from_size = self.source._file_size(from_entry, from_stat)
690
to_size = self.target._file_size(to_entry, to_stat)
691
if from_size != to_size:
692
changed_content = True
693
elif (self.source.get_file_sha1(file_id, from_path, from_stat) !=
694
self.target.get_file_sha1(file_id, to_path, to_stat)):
695
changed_content = True
696
elif from_kind == 'symlink':
697
if (self.source.get_symlink_target(file_id) !=
698
self.target.get_symlink_target(file_id)):
699
changed_content = True
700
elif from_kind == 'tree-reference':
701
if (self.source.get_reference_revision(file_id, from_path)
702
!= self.target.get_reference_revision(file_id, to_path)):
703
changed_content = True
704
parent = (from_parent, to_entry.parent_id)
705
name = (from_name, to_entry.name)
706
executable = (from_executable, to_executable)
708
pb.update('comparing files', entry_count, num_entries)
709
if (changed_content is not False or versioned[0] != versioned[1]
710
or parent[0] != parent[1] or name[0] != name[1] or
711
executable[0] != executable[1] or include_unchanged):
712
yield (file_id, (from_path, to_path), changed_content,
713
versioned, parent, name, kind, executable)
715
while all_unversioned:
716
# yield any trailing unversioned paths
717
unversioned_path = all_unversioned.popleft()
718
to_kind, to_executable, to_stat = \
719
self.target._comparison_data(fake_entry, unversioned_path[1])
720
yield (None, (None, unversioned_path[1]), True, (False, False),
722
(None, unversioned_path[0][-1]),
724
(None, to_executable))
726
def get_to_path(to_entry):
727
if to_entry.parent_id is None:
728
to_path = '' # the root
730
if to_entry.parent_id not in to_paths:
732
return get_to_path(self.target.inventory[to_entry.parent_id])
733
to_path = osutils.pathjoin(to_paths[to_entry.parent_id],
735
to_paths[to_entry.file_id] = to_path
738
for path, from_entry in from_entries_by_dir:
739
file_id = from_entry.file_id
740
if file_id in to_paths:
743
if not file_id in self.target.inventory:
744
# common case - paths we have not emitted are not present in
748
to_path = get_to_path(self.target.inventory[file_id])
751
pb.update('comparing files', entry_count, num_entries)
752
versioned = (True, False)
753
parent = (from_entry.parent_id, None)
754
name = (from_entry.name, None)
755
from_kind, from_executable, stat_value = \
756
self.source._comparison_data(from_entry, path)
757
kind = (from_kind, None)
758
executable = (from_executable, None)
759
changed_content = True
760
# the parent's path is necessarily known at this point.
761
yield(file_id, (path, to_path), changed_content, versioned, parent,
762
name, kind, executable)
765
# This was deprecated before 0.12, but did not have an official warning
766
@symbol_versioning.deprecated_function(symbol_versioning.zero_twelve)
767
def RevisionTree(*args, **kwargs):
768
"""RevisionTree has moved to bzrlib.revisiontree.RevisionTree()
770
Accessing it as bzrlib.tree.RevisionTree has been deprecated as of
773
from bzrlib.revisiontree import RevisionTree as _RevisionTree
774
return _RevisionTree(*args, **kwargs)
added
removed
bzrlib/tree.py
