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"])
269
def path2id(self, path):
270
"""Return the id for path in this tree."""
271
return self._inventory.path2id(path)
273
def paths2ids(self, paths, trees=[], require_versioned=True):
274
"""Return all the ids that can be reached by walking from paths.
276
Each path is looked up in each this tree and any extras provided in
277
trees, and this is repeated recursively: the children in an extra tree
278
of a directory that has been renamed under a provided path in this tree
279
are all returned, even if none exist until a provided path in this
280
tree, and vice versa.
282
:param paths: An iterable of paths to start converting to ids from.
283
Alternatively, if paths is None, no ids should be calculated and None
284
will be returned. This is offered to make calling the api unconditional
285
for code that *might* take a list of files.
286
:param trees: Additional trees to consider.
287
:param require_versioned: If False, do not raise NotVersionedError if
288
an element of paths is not versioned in this tree and all of trees.
290
return find_ids_across_trees(paths, [self] + list(trees), require_versioned)
292
def print_file(self, file_id):
293
"""Print file with id `file_id` to stdout."""
294
file_id = osutils.safe_file_id(file_id)
296
sys.stdout.write(self.get_file_text(file_id))
301
def revision_tree(self, revision_id):
302
"""Obtain a revision tree for the revision revision_id.
304
The intention of this method is to allow access to possibly cached
305
tree data. Implementors of this method should raise NoSuchRevision if
306
the tree is not locally available, even if they could obtain the
307
tree via a repository or some other means. Callers are responsible
308
for finding the ultimate source for a revision tree.
310
:param revision_id: The revision_id of the requested tree.
312
:raises: NoSuchRevision if the tree cannot be obtained.
314
raise errors.NoSuchRevisionInTree(self, revision_id)
317
"""What files are present in this tree and unknown.
319
:return: an iterator over the unknown files.
326
def filter_unversioned_files(self, paths):
327
"""Filter out paths that are versioned.
329
:return: set of paths.
331
# NB: we specifically *don't* call self.has_filename, because for
332
# WorkingTrees that can indicate files that exist on disk but that
334
pred = self.inventory.has_filename
335
return set((p for p in paths if not pred(p)))
337
def walkdirs(self, prefix=""):
338
"""Walk the contents of this tree from path down.
340
This yields all the data about the contents of a directory at a time.
341
After each directory has been yielded, if the caller has mutated the
342
list to exclude some directories, they are then not descended into.
344
The data yielded is of the form:
345
((directory-relpath, directory-path-from-root, directory-fileid),
346
[(relpath, basename, kind, lstat, path_from_tree_root, file_id,
347
versioned_kind), ...]),
348
- directory-relpath is the containing dirs relpath from prefix
349
- directory-path-from-root is the containing dirs path from /
350
- directory-fileid is the id of the directory if it is versioned.
351
- relpath is the relative path within the subtree being walked.
352
- basename is the basename
353
- kind is the kind of the file now. If unknonwn then the file is not
354
present within the tree - but it may be recorded as versioned. See
356
- lstat is the stat data *if* the file was statted.
357
- path_from_tree_root is the path from the root of the tree.
358
- file_id is the file_id is the entry is versioned.
359
- versioned_kind is the kind of the file as last recorded in the
360
versioning system. If 'unknown' the file is not versioned.
361
One of 'kind' and 'versioned_kind' must not be 'unknown'.
363
:param prefix: Start walking from prefix within the tree rather than
364
at the root. This allows one to walk a subtree but get paths that are
365
relative to a tree rooted higher up.
366
:return: an iterator over the directory data.
368
raise NotImplementedError(self.walkdirs)
371
class EmptyTree(Tree):
374
self._inventory = Inventory(root_id=None)
375
symbol_versioning.warn('EmptyTree is deprecated as of bzr 0.9 please'
376
' use repository.revision_tree instead.',
377
DeprecationWarning, stacklevel=2)
379
def get_parent_ids(self):
382
def get_symlink_target(self, file_id):
385
def has_filename(self, filename):
388
def kind(self, file_id):
389
file_id = osutils.safe_file_id(file_id)
390
assert self._inventory[file_id].kind == "directory"
393
def list_files(self, include_root=False):
396
def __contains__(self, file_id):
397
file_id = osutils.safe_file_id(file_id)
398
return (file_id in self._inventory)
400
def get_file_sha1(self, file_id, path=None, stat_value=None):
404
######################################################################
407
# TODO: Merge these two functions into a single one that can operate
408
# on either a whole tree or a set of files.
410
# TODO: Return the diff in order by filename, not by category or in
411
# random order. Can probably be done by lock-stepping through the
412
# filenames from both trees.
415
def file_status(filename, old_tree, new_tree):
416
"""Return single-letter status, old and new names for a file.
418
The complexity here is in deciding how to represent renames;
419
many complex cases are possible.
421
old_inv = old_tree.inventory
422
new_inv = new_tree.inventory
423
new_id = new_inv.path2id(filename)
424
old_id = old_inv.path2id(filename)
426
if not new_id and not old_id:
427
# easy: doesn't exist in either; not versioned at all
428
if new_tree.is_ignored(filename):
429
return 'I', None, None
431
return '?', None, None
433
# There is now a file of this name, great.
436
# There is no longer a file of this name, but we can describe
437
# what happened to the file that used to have
438
# this name. There are two possibilities: either it was
439
# deleted entirely, or renamed.
441
if new_inv.has_id(old_id):
442
return 'X', old_inv.id2path(old_id), new_inv.id2path(old_id)
444
return 'D', old_inv.id2path(old_id), None
446
# if the file_id is new in this revision, it is added
447
if new_id and not old_inv.has_id(new_id):
450
# if there used to be a file of this name, but that ID has now
451
# disappeared, it is deleted
452
if old_id and not new_inv.has_id(old_id):
459
def find_renames(old_inv, new_inv):
460
for file_id in old_inv:
461
if file_id not in new_inv:
463
old_name = old_inv.id2path(file_id)
464
new_name = new_inv.id2path(file_id)
465
if old_name != new_name:
466
yield (old_name, new_name)
469
def find_ids_across_trees(filenames, trees, require_versioned=True):
470
"""Find the ids corresponding to specified filenames.
472
All matches in all trees will be used, and all children of matched
473
directories will be used.
475
:param filenames: The filenames to find file_ids for (if None, returns
477
:param trees: The trees to find file_ids within
478
:param require_versioned: if true, all specified filenames must occur in
480
:return: a set of file ids for the specified filenames and their children.
484
specified_path_ids = _find_ids_across_trees(filenames, trees,
486
return _find_children_across_trees(specified_path_ids, trees)
489
def _find_ids_across_trees(filenames, trees, require_versioned):
490
"""Find the ids corresponding to specified filenames.
492
All matches in all trees will be used, but subdirectories are not scanned.
494
:param filenames: The filenames to find file_ids for
495
:param trees: The trees to find file_ids within
496
:param require_versioned: if true, all specified filenames must occur in
498
:return: a set of (path, file ids) for the specified filenames
501
interesting_ids = set()
502
for tree_path in filenames:
505
file_id = tree.path2id(tree_path)
506
if file_id is not None:
507
interesting_ids.add(file_id)
510
not_versioned.append(tree_path)
511
if len(not_versioned) > 0 and require_versioned:
512
raise errors.PathsNotVersionedError(not_versioned)
513
return interesting_ids
516
def _find_children_across_trees(specified_ids, trees):
517
"""Return a set including specified ids and their children
519
All matches in all trees will be used.
521
:param trees: The trees to find file_ids within
522
:return: a set containing all specified ids and their children
524
interesting_ids = set(specified_ids)
525
pending = interesting_ids
526
# now handle children of interesting ids
527
# we loop so that we handle all children of each id in both trees
528
while len(pending) > 0:
530
for file_id in pending:
532
if not tree.has_id(file_id):
534
entry = tree.inventory[file_id]
535
for child in getattr(entry, 'children', {}).itervalues():
536
if child.file_id not in interesting_ids:
537
new_pending.add(child.file_id)
538
interesting_ids.update(new_pending)
539
pending = new_pending
540
return interesting_ids
543
class InterTree(InterObject):
544
"""This class represents operations taking place between two Trees.
546
Its instances have methods like 'compare' and contain references to the
547
source and target trees these operations are to be carried out on.
549
clients of bzrlib should not need to use InterTree directly, rather they
550
should use the convenience methods on Tree such as 'Tree.compare()' which
551
will pass through to InterTree as appropriate.
557
def compare(self, want_unchanged=False, specific_files=None,
558
extra_trees=None, require_versioned=False, include_root=False,
559
want_unversioned=False):
560
"""Return the changes from source to target.
562
:return: A TreeDelta.
563
:param specific_files: An optional list of file paths to restrict the
564
comparison to. When mapping filenames to ids, all matches in all
565
trees (including optional extra_trees) are used, and all children of
566
matched directories are included.
567
:param want_unchanged: An optional boolean requesting the inclusion of
568
unchanged entries in the result.
569
:param extra_trees: An optional list of additional trees to use when
570
mapping the contents of specific_files (paths) to file_ids.
571
:param require_versioned: An optional boolean (defaults to False). When
572
supplied and True all the 'specific_files' must be versioned, or
573
a PathsNotVersionedError will be thrown.
574
:param want_unversioned: Scan for unversioned paths.
576
# NB: show_status depends on being able to pass in non-versioned files
577
# and report them as unknown
578
trees = (self.source,)
579
if extra_trees is not None:
580
trees = trees + tuple(extra_trees)
581
# target is usually the newer tree:
582
specific_file_ids = self.target.paths2ids(specific_files, trees,
583
require_versioned=require_versioned)
584
if specific_files and not specific_file_ids:
585
# All files are unversioned, so just return an empty delta
586
# _compare_trees would think we want a complete delta
587
result = delta.TreeDelta()
588
fake_entry = InventoryFile('unused', 'unused', 'unused')
589
result.unversioned = [(path, None,
590
self.target._comparison_data(fake_entry, path)[0]) for path in
593
return delta._compare_trees(self.source, self.target, want_unchanged,
594
specific_files, include_root, extra_trees=extra_trees,
595
want_unversioned=want_unversioned)
597
def _iter_changes(self, include_unchanged=False,
598
specific_files=None, pb=None, extra_trees=[],
599
require_versioned=True, want_unversioned=False):
600
"""Generate an iterator of changes between trees.
603
(file_id, (path_in_source, path_in_target),
604
changed_content, versioned, parent, name, kind,
607
Changed_content is True if the file's content has changed. This
608
includes changes to its kind, and to a symlink's target.
610
versioned, parent, name, kind, executable are tuples of (from, to).
611
If a file is missing in a tree, its kind is None.
613
Iteration is done in parent-to-child order, relative to the target
616
There is no guarantee that all paths are in sorted order: the
617
requirement to expand the search due to renames may result in children
618
that should be found early being found late in the search, after
619
lexically later results have been returned.
620
:param require_versioned: Raise errors.PathsNotVersionedError if a
621
path in the specific_files list is not versioned in one of
622
source, target or extra_trees.
623
:param want_unversioned: Should unversioned files be returned in the
624
output. An unversioned file is defined as one with (False, False)
625
for the versioned pair.
628
lookup_trees = [self.source]
630
lookup_trees.extend(extra_trees)
631
specific_file_ids = self.target.paths2ids(specific_files,
632
lookup_trees, require_versioned=require_versioned)
634
all_unversioned = sorted([(p.split('/'), p) for p in self.target.extras()
635
if not specific_files or
636
osutils.is_inside_any(specific_files, p)])
637
all_unversioned = deque(all_unversioned)
639
all_unversioned = deque()
641
from_entries_by_dir = list(self.source.inventory.iter_entries_by_dir(
642
specific_file_ids=specific_file_ids))
643
from_data = dict((e.file_id, (p, e)) for p, e in from_entries_by_dir)
644
to_entries_by_dir = list(self.target.inventory.iter_entries_by_dir(
645
specific_file_ids=specific_file_ids))
646
num_entries = len(from_entries_by_dir) + len(to_entries_by_dir)
648
# the unversioned path lookup only occurs on real trees - where there
649
# can be extras. So the fake_entry is solely used to look up
650
# executable it values when execute is not supported.
651
fake_entry = InventoryFile('unused', 'unused', 'unused')
652
for to_path, to_entry in to_entries_by_dir:
653
while all_unversioned and all_unversioned[0][0] < to_path.split('/'):
654
unversioned_path = all_unversioned.popleft()
655
to_kind, to_executable, to_stat = \
656
self.target._comparison_data(fake_entry, unversioned_path[1])
657
yield (None, (None, unversioned_path[1]), True, (False, False),
659
(None, unversioned_path[0][-1]),
661
(None, to_executable))
662
file_id = to_entry.file_id
663
to_paths[file_id] = to_path
665
changed_content = False
666
from_path, from_entry = from_data.get(file_id, (None, None))
667
from_versioned = (from_entry is not None)
668
if from_entry is not None:
669
from_versioned = True
670
from_name = from_entry.name
671
from_parent = from_entry.parent_id
672
from_kind, from_executable, from_stat = \
673
self.source._comparison_data(from_entry, from_path)
676
from_versioned = False
680
from_executable = None
681
versioned = (from_versioned, True)
682
to_kind, to_executable, to_stat = \
683
self.target._comparison_data(to_entry, to_path)
684
kind = (from_kind, to_kind)
685
if kind[0] != kind[1]:
686
changed_content = True
687
elif from_kind == 'file':
688
from_size = self.source._file_size(from_entry, from_stat)
689
to_size = self.target._file_size(to_entry, to_stat)
690
if from_size != to_size:
691
changed_content = True
692
elif (self.source.get_file_sha1(file_id, from_path, from_stat) !=
693
self.target.get_file_sha1(file_id, to_path, to_stat)):
694
changed_content = True
695
elif from_kind == 'symlink':
696
if (self.source.get_symlink_target(file_id) !=
697
self.target.get_symlink_target(file_id)):
698
changed_content = True
699
elif from_kind == 'tree-reference':
700
if (self.source.get_reference_revision(file_id, from_path)
701
!= self.target.get_reference_revision(file_id, to_path)):
702
changed_content = True
703
parent = (from_parent, to_entry.parent_id)
704
name = (from_name, to_entry.name)
705
executable = (from_executable, to_executable)
707
pb.update('comparing files', entry_count, num_entries)
708
if (changed_content is not False or versioned[0] != versioned[1]
709
or parent[0] != parent[1] or name[0] != name[1] or
710
executable[0] != executable[1] or include_unchanged):
711
yield (file_id, (from_path, to_path), changed_content,
712
versioned, parent, name, kind, executable)
714
while all_unversioned:
715
# yield any trailing unversioned paths
716
unversioned_path = all_unversioned.popleft()
717
to_kind, to_executable, to_stat = \
718
self.target._comparison_data(fake_entry, unversioned_path[1])
719
yield (None, (None, unversioned_path[1]), True, (False, False),
721
(None, unversioned_path[0][-1]),
723
(None, to_executable))
725
def get_to_path(to_entry):
726
if to_entry.parent_id is None:
727
to_path = '' # the root
729
if to_entry.parent_id not in to_paths:
731
return get_to_path(self.target.inventory[to_entry.parent_id])
732
to_path = osutils.pathjoin(to_paths[to_entry.parent_id],
734
to_paths[to_entry.file_id] = to_path
737
for path, from_entry in from_entries_by_dir:
738
file_id = from_entry.file_id
739
if file_id in to_paths:
742
if not file_id in self.target.inventory:
743
# common case - paths we have not emitted are not present in
747
to_path = get_to_path(self.target.inventory[file_id])
750
pb.update('comparing files', entry_count, num_entries)
751
versioned = (True, False)
752
parent = (from_entry.parent_id, None)
753
name = (from_entry.name, None)
754
from_kind, from_executable, stat_value = \
755
self.source._comparison_data(from_entry, path)
756
kind = (from_kind, None)
757
executable = (from_executable, None)
758
changed_content = True
759
# the parent's path is necessarily known at this point.
760
yield(file_id, (path, to_path), changed_content, versioned, parent,
761
name, kind, executable)
764
# This was deprecated before 0.12, but did not have an official warning
765
@symbol_versioning.deprecated_function(symbol_versioning.zero_twelve)
766
def RevisionTree(*args, **kwargs):
767
"""RevisionTree has moved to bzrlib.revisiontree.RevisionTree()
769
Accessing it as bzrlib.tree.RevisionTree has been deprecated as of
772
from bzrlib.revisiontree import RevisionTree as _RevisionTree
773
return _RevisionTree(*args, **kwargs)
added
removed
bzrlib/tree.py
