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 cStringIO import StringIO
29
from bzrlib.decorators import needs_read_lock
30
from bzrlib.errors import BzrError, BzrCheckError
31
from bzrlib import errors
32
from bzrlib.inventory import Inventory
33
from bzrlib.inter import InterObject
34
from bzrlib.osutils import fingerprint_file
35
import bzrlib.revision
36
from bzrlib.trace import mutter, note
40
"""Abstract file tree.
42
There are several subclasses:
44
* `WorkingTree` exists as files on disk editable by the user.
46
* `RevisionTree` is a tree as recorded at some point in the past.
48
Trees contain an `Inventory` object, and also know how to retrieve
49
file texts mentioned in the inventory, either from a working
50
directory or from a store.
52
It is possible for trees to contain files that are not described
53
in their inventory or vice versa; for this use `filenames()`.
55
Trees can be compared, etc, regardless of whether they are working
56
trees or versioned trees.
59
def changes_from(self, other, want_unchanged=False, specific_files=None,
60
extra_trees=None, require_versioned=False, include_root=False):
61
"""Return a TreeDelta of the changes from other to this tree.
63
:param other: A tree to compare with.
64
:param specific_files: An optional list of file paths to restrict the
65
comparison to. When mapping filenames to ids, all matches in all
66
trees (including optional extra_trees) are used, and all children of
67
matched directories are included.
68
:param want_unchanged: An optional boolean requesting the inclusion of
69
unchanged entries in the result.
70
:param extra_trees: An optional list of additional trees to use when
71
mapping the contents of specific_files (paths) to file_ids.
72
:param require_versioned: An optional boolean (defaults to False). When
73
supplied and True all the 'specific_files' must be versioned, or
74
a PathsNotVersionedError will be thrown.
76
The comparison will be performed by an InterTree object looked up on
79
# Martin observes that Tree.changes_from returns a TreeDelta and this
80
# may confuse people, because the class name of the returned object is
81
# a synonym of the object referenced in the method name.
82
return InterTree.get(other, self).compare(
83
want_unchanged=want_unchanged,
84
specific_files=specific_files,
85
extra_trees=extra_trees,
86
require_versioned=require_versioned,
87
include_root=include_root
90
def _iter_changes(self, from_tree, include_unchanged=False,
91
specific_files=None, pb=None, extra_trees=None,
92
require_versioned=True):
93
intertree = InterTree.get(from_tree, self)
94
return intertree._iter_changes(include_unchanged, specific_files, pb,
95
extra_trees, require_versioned)
98
"""Get a list of the conflicts in the tree.
100
Each conflict is an instance of bzrlib.conflicts.Conflict.
104
def get_parent_ids(self):
105
"""Get the parent ids for this tree.
107
:return: a list of parent ids. [] is returned to indicate
108
a tree with no parents.
109
:raises: BzrError if the parents are not known.
111
raise NotImplementedError(self.get_parent_ids)
113
def has_filename(self, filename):
114
"""True if the tree has given filename."""
115
raise NotImplementedError()
117
def has_id(self, file_id):
118
file_id = osutils.safe_file_id(file_id)
119
return self.inventory.has_id(file_id)
121
__contains__ = has_id
123
def has_or_had_id(self, file_id):
124
file_id = osutils.safe_file_id(file_id)
125
if file_id == self.inventory.root.file_id:
127
return self.inventory.has_id(file_id)
130
return iter(self.inventory)
132
def id2path(self, file_id):
133
file_id = osutils.safe_file_id(file_id)
134
return self.inventory.id2path(file_id)
136
def is_control_filename(self, filename):
137
"""True if filename is the name of a control file in this tree.
139
:param filename: A filename within the tree. This is a relative path
140
from the root of this tree.
142
This is true IF and ONLY IF the filename is part of the meta data
143
that bzr controls in this tree. I.E. a random .bzr directory placed
144
on disk will not be a control file for this tree.
146
return self.bzrdir.is_control_filename(filename)
149
def iter_entries_by_dir(self, specific_file_ids=None):
150
"""Walk the tree in 'by_dir' order.
152
This will yield each entry in the tree as a (path, entry) tuple. The
153
order that they are yielded is: the contents of a directory are
154
preceeded by the parent of a directory, and all the contents of a
155
directory are grouped together.
157
return self.inventory.iter_entries_by_dir(
158
specific_file_ids=specific_file_ids)
160
def kind(self, file_id):
161
raise NotImplementedError("subclasses must implement kind")
163
def _comparison_data(self, entry, path):
164
"""Return a tuple of kind, executable, stat_value for a file.
166
entry may be None if there is no inventory entry for the file, but
167
path must always be supplied.
169
kind is None if there is no file present (even if an inventory id is
170
present). executable is False for non-file entries.
172
raise NotImplementedError(self._comparison_data)
174
def _file_size(self, entry, stat_value):
175
raise NotImplementedError(self._file_size)
177
def _get_inventory(self):
178
return self._inventory
180
def get_file(self, file_id):
181
"""Return a file object for the file file_id in the tree."""
182
raise NotImplementedError(self.get_file)
184
def get_file_by_path(self, path):
185
return self.get_file(self._inventory.path2id(path))
187
def get_symlink_target(self, file_id):
188
"""Get the target for a given file_id.
190
It is assumed that the caller already knows that file_id is referencing
192
:param file_id: Handle for the symlink entry.
193
:return: The path the symlink points to.
195
raise NotImplementedError(self.get_symlink_target)
197
def annotate_iter(self, file_id):
198
"""Return an iterator of revision_id, line tuples
200
For working trees (and mutable trees in general), the special
201
revision_id 'current:' will be used for lines that are new in this
202
tree, e.g. uncommitted changes.
203
:param file_id: The file to produce an annotated version from
205
raise NotImplementedError(self.annotate_iter)
207
inventory = property(_get_inventory,
208
doc="Inventory of this Tree")
210
def _check_retrieved(self, ie, f):
213
fp = fingerprint_file(f)
216
if ie.text_size is not None:
217
if ie.text_size != fp['size']:
218
raise BzrError("mismatched size for file %r in %r" % (ie.file_id, self._store),
219
["inventory expects %d bytes" % ie.text_size,
220
"file is actually %d bytes" % fp['size'],
221
"store is probably damaged/corrupt"])
223
if ie.text_sha1 != fp['sha1']:
224
raise BzrError("wrong SHA-1 for file %r in %r" % (ie.file_id, self._store),
225
["inventory expects %s" % ie.text_sha1,
226
"file is actually %s" % fp['sha1'],
227
"store is probably damaged/corrupt"])
229
def path2id(self, path):
230
"""Return the id for path in this tree."""
231
return self._inventory.path2id(path)
233
def paths2ids(self, paths, trees=[], require_versioned=True):
234
"""Return all the ids that can be reached by walking from paths.
236
Each path is looked up in each this tree and any extras provided in
237
trees, and this is repeated recursively: the children in an extra tree
238
of a directory that has been renamed under a provided path in this tree
239
are all returned, even if none exist until a provided path in this
240
tree, and vice versa.
242
:param paths: An iterable of paths to start converting to ids from.
243
Alternatively, if paths is None, no ids should be calculated and None
244
will be returned. This is offered to make calling the api unconditional
245
for code that *might* take a list of files.
246
:param trees: Additional trees to consider.
247
:param require_versioned: If False, do not raise NotVersionedError if
248
an element of paths is not versioned in this tree and all of trees.
250
return find_ids_across_trees(paths, [self] + list(trees), require_versioned)
252
def print_file(self, file_id):
253
"""Print file with id `file_id` to stdout."""
254
file_id = osutils.safe_file_id(file_id)
256
sys.stdout.write(self.get_file_text(file_id))
261
def revision_tree(self, revision_id):
262
"""Obtain a revision tree for the revision revision_id.
264
The intention of this method is to allow access to possibly cached
265
tree data. Implementors of this method should raise NoSuchRevision if
266
the tree is not locally available, even if they could obtain the
267
tree via a repository or some other means. Callers are responsible
268
for finding the ultimate source for a revision tree.
270
:param revision_id: The revision_id of the requested tree.
272
:raises: NoSuchRevision if the tree cannot be obtained.
274
raise errors.NoSuchRevisionInTree(self, revision_id)
277
"""What files are present in this tree and unknown.
279
:return: an iterator over the unknown files.
286
def filter_unversioned_files(self, paths):
287
"""Filter out paths that are not versioned.
289
:return: set of paths.
291
# NB: we specifically *don't* call self.has_filename, because for
292
# WorkingTrees that can indicate files that exist on disk but that
294
pred = self.inventory.has_filename
295
return set((p for p in paths if not pred(p)))
297
def walkdirs(self, prefix=""):
298
"""Walk the contents of this tree from path down.
300
This yields all the data about the contents of a directory at a time.
301
After each directory has been yielded, if the caller has mutated the
302
list to exclude some directories, they are then not descended into.
304
The data yielded is of the form:
305
((directory-relpath, directory-path-from-root, directory-fileid),
306
[(relpath, basename, kind, lstat, path_from_tree_root, file_id,
307
versioned_kind), ...]),
308
- directory-relpath is the containing dirs relpath from prefix
309
- directory-path-from-root is the containing dirs path from /
310
- directory-fileid is the id of the directory if it is versioned.
311
- relpath is the relative path within the subtree being walked.
312
- basename is the basename
313
- kind is the kind of the file now. If unknonwn then the file is not
314
present within the tree - but it may be recorded as versioned. See
316
- lstat is the stat data *if* the file was statted.
317
- path_from_tree_root is the path from the root of the tree.
318
- file_id is the file_id is the entry is versioned.
319
- versioned_kind is the kind of the file as last recorded in the
320
versioning system. If 'unknown' the file is not versioned.
321
One of 'kind' and 'versioned_kind' must not be 'unknown'.
323
:param prefix: Start walking from prefix within the tree rather than
324
at the root. This allows one to walk a subtree but get paths that are
325
relative to a tree rooted higher up.
326
:return: an iterator over the directory data.
328
raise NotImplementedError(self.walkdirs)
331
class EmptyTree(Tree):
334
self._inventory = Inventory(root_id=None)
335
symbol_versioning.warn('EmptyTree is deprecated as of bzr 0.9 please'
336
' use repository.revision_tree instead.',
337
DeprecationWarning, stacklevel=2)
339
def get_parent_ids(self):
342
def get_symlink_target(self, file_id):
345
def has_filename(self, filename):
348
def kind(self, file_id):
349
file_id = osutils.safe_file_id(file_id)
350
assert self._inventory[file_id].kind == "directory"
353
def list_files(self, include_root=False):
356
def __contains__(self, file_id):
357
file_id = osutils.safe_file_id(file_id)
358
return (file_id in self._inventory)
360
def get_file_sha1(self, file_id, path=None, stat_value=None):
364
######################################################################
367
# TODO: Merge these two functions into a single one that can operate
368
# on either a whole tree or a set of files.
370
# TODO: Return the diff in order by filename, not by category or in
371
# random order. Can probably be done by lock-stepping through the
372
# filenames from both trees.
375
def file_status(filename, old_tree, new_tree):
376
"""Return single-letter status, old and new names for a file.
378
The complexity here is in deciding how to represent renames;
379
many complex cases are possible.
381
old_inv = old_tree.inventory
382
new_inv = new_tree.inventory
383
new_id = new_inv.path2id(filename)
384
old_id = old_inv.path2id(filename)
386
if not new_id and not old_id:
387
# easy: doesn't exist in either; not versioned at all
388
if new_tree.is_ignored(filename):
389
return 'I', None, None
391
return '?', None, None
393
# There is now a file of this name, great.
396
# There is no longer a file of this name, but we can describe
397
# what happened to the file that used to have
398
# this name. There are two possibilities: either it was
399
# deleted entirely, or renamed.
401
if new_inv.has_id(old_id):
402
return 'X', old_inv.id2path(old_id), new_inv.id2path(old_id)
404
return 'D', old_inv.id2path(old_id), None
406
# if the file_id is new in this revision, it is added
407
if new_id and not old_inv.has_id(new_id):
410
# if there used to be a file of this name, but that ID has now
411
# disappeared, it is deleted
412
if old_id and not new_inv.has_id(old_id):
419
def find_renames(old_inv, new_inv):
420
for file_id in old_inv:
421
if file_id not in new_inv:
423
old_name = old_inv.id2path(file_id)
424
new_name = new_inv.id2path(file_id)
425
if old_name != new_name:
426
yield (old_name, new_name)
429
def find_ids_across_trees(filenames, trees, require_versioned=True):
430
"""Find the ids corresponding to specified filenames.
432
All matches in all trees will be used, and all children of matched
433
directories will be used.
435
:param filenames: The filenames to find file_ids for (if None, returns
437
:param trees: The trees to find file_ids within
438
:param require_versioned: if true, all specified filenames must occur in
440
:return: a set of file ids for the specified filenames and their children.
444
specified_path_ids = _find_ids_across_trees(filenames, trees,
446
return _find_children_across_trees(specified_path_ids, trees)
447
# specified_ids = [id for path, id in _find_path_ids_across_trees(filenames, trees, require_versioned)]
448
# return _find_children_across_trees(specified_ids, trees)
450
def find_path_ids_across_trees(filenames, trees, require_versioned=True):
451
"""Find the paths and ids corresponding to specified filenames.
453
All matches in all trees will be used, and all children of matched
454
directories will be included
456
:param filenames: The filenames to find file_ids for
457
:param trees: The trees to find file_ids within
458
:param require_versioned: if true, all specified filenames must occur in
460
:return: a set of (path, file ids) for the specified filenames and their
461
children. The returned path is the path of the id in the first tree
462
that contains it. This matters when files have been moved
466
# This function needs to know the ids for filenames in all trees, then
467
# search for those same files and children in all the other trees.
468
# it is complicated by the same path in two trees being able to have
469
# different ids, which might both be present in both trees.
470
# consider two trees, which have had 'mv foo bar' and 'mv baz foo' done
471
# in this case, a diff of 'foo' should should changes to both the current
472
# 'bar' and the current 'foo' which was baz. Its arguable that if
473
# the situation is 'mv parent/foo bar' and 'mv baz parent/foo', that
474
# we should return the current bar and the current parent/foo' - at the
475
# moment we do, but we loop around all ids and all trees: I*T checks.
477
# Updating this algorithm to be fast in the common case:
478
# nothing has moved, all files have the same id in parent, child and there
479
# are only two trees (or one is working tree and the others are parents).
480
# walk the dirstate. as we find each path, gather the paths of that
481
# id in all trees. add a mapping from the id to the path in those trees.
482
# now lookup children by id, again in all trees; for these trees that
483
# nothing has moved in, the id->path mapping will allow us to find the
484
# parent trivially. To answer 'has anything been moved' in one of the
485
# dirstate parent trees though, we will need to stare harder at it.
487
# Now, given a path index, that is trivial for any one tree, and given
488
# that we can ask for additional data from a dirstate tree, its a single
489
# pass, though it will require scanning the entire tree to find paths
490
# that were at the current location.
491
# ideal results?: There are three things: tree, path, id. Pathologically
492
# we can have completely disjoint ids for each tree; but we cannot have
493
# disjoin paths for each tree, except if we scan each tree for the
494
# different ids from other trees.
496
specified_path_ids = _find_ids_across_trees(filenames, trees,
498
return _find_path_id_children_across_trees(specified_path_ids, trees)
501
def _find_ids_across_trees(filenames, trees, require_versioned):
502
"""Find the ids corresponding to specified filenames.
504
All matches in all trees will be used, but subdirectories are not scanned.
506
:param filenames: The filenames to find file_ids for
507
:param trees: The trees to find file_ids within
508
:param require_versioned: if true, all specified filenames must occur in
510
:return: a set of (path, file ids) for the specified filenames
513
interesting_ids = set()
514
for tree_path in filenames:
517
file_id = tree.path2id(tree_path)
518
if file_id is not None:
519
interesting_ids.add(file_id)
522
not_versioned.append(tree_path)
523
if len(not_versioned) > 0 and require_versioned:
524
raise errors.PathsNotVersionedError(not_versioned)
525
return interesting_ids
528
def _find_children_across_trees(specified_ids, trees):
529
"""Return a set including specified ids and their children
531
All matches in all trees will be used.
533
:param trees: The trees to find file_ids within
534
:return: a set containing all specified ids and their children
536
interesting_ids = set(specified_ids)
537
pending = interesting_ids
538
# now handle children of interesting ids
539
# we loop so that we handle all children of each id in both trees
540
while len(pending) > 0:
542
for file_id in pending:
544
if not tree.has_id(file_id):
546
entry = tree.inventory[file_id]
547
for child in getattr(entry, 'children', {}).itervalues():
548
if child.file_id not in interesting_ids:
549
new_pending.add(child.file_id)
550
interesting_ids.update(new_pending)
551
pending = new_pending
552
return interesting_ids
555
class InterTree(InterObject):
556
"""This class represents operations taking place between two Trees.
558
Its instances have methods like 'compare' and contain references to the
559
source and target trees these operations are to be carried out on.
561
clients of bzrlib should not need to use InterTree directly, rather they
562
should use the convenience methods on Tree such as 'Tree.compare()' which
563
will pass through to InterTree as appropriate.
569
def compare(self, want_unchanged=False, specific_files=None,
570
extra_trees=None, require_versioned=False, include_root=False):
571
"""Return the changes from source to target.
573
:return: A TreeDelta.
574
:param specific_files: An optional list of file paths to restrict the
575
comparison to. When mapping filenames to ids, all matches in all
576
trees (including optional extra_trees) are used, and all children of
577
matched directories are included.
578
:param want_unchanged: An optional boolean requesting the inclusion of
579
unchanged entries in the result.
580
:param extra_trees: An optional list of additional trees to use when
581
mapping the contents of specific_files (paths) to file_ids.
582
:param require_versioned: An optional boolean (defaults to False). When
583
supplied and True all the 'specific_files' must be versioned, or
584
a PathsNotVersionedError will be thrown.
586
# NB: show_status depends on being able to pass in non-versioned files
587
# and report them as unknown
588
trees = (self.source,)
589
if extra_trees is not None:
590
trees = trees + tuple(extra_trees)
591
# target is usually the newer tree:
592
specific_file_ids = self.target.paths2ids(specific_files, trees,
593
require_versioned=require_versioned)
594
if specific_files and not specific_file_ids:
595
# All files are unversioned, so just return an empty delta
596
# _compare_trees would think we want a complete delta
597
return delta.TreeDelta()
598
return delta._compare_trees(self.source, self.target, want_unchanged,
599
specific_files, include_root, extra_trees=extra_trees)
601
def _iter_changes(self, include_unchanged=False,
602
specific_files=None, pb=None, extra_trees=[],
603
require_versioned=True):
604
"""Generate an iterator of changes between trees.
607
(file_id, path, changed_content, versioned, parent, name, kind,
610
Path is relative to the target tree. changed_content is True if the
611
file's content has changed. This includes changes to its kind, and to
614
versioned, parent, name, kind, executable are tuples of (from, to).
615
If a file is missing in a tree, its kind is None.
617
Iteration is done in parent-to-child order, relative to the target
620
There is no guarantee that all paths are in sorted order: the
621
requirement to expand the search due to renames may result in children
622
that should be found early being found late in the search, after
623
lexically later results have been returned.
624
:param require_versioned: Raise errors.PathsNotVersionedError if a
625
path in the specific_files list is not versioned in one of
626
source, target or extra_trees.
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
from_entries_by_dir = list(self.source.inventory.iter_entries_by_dir(
635
specific_file_ids=specific_file_ids))
636
from_data = dict((e.file_id, (p, e)) for p, e in from_entries_by_dir)
637
to_entries_by_dir = list(self.target.inventory.iter_entries_by_dir(
638
specific_file_ids=specific_file_ids))
639
num_entries = len(from_entries_by_dir) + len(to_entries_by_dir)
641
for to_path, to_entry in to_entries_by_dir:
642
file_id = to_entry.file_id
643
to_paths[file_id] = to_path
645
changed_content = False
646
from_path, from_entry = from_data.get(file_id, (None, None))
647
from_versioned = (from_entry is not None)
648
if from_entry is not None:
649
from_versioned = True
650
from_name = from_entry.name
651
from_parent = from_entry.parent_id
652
from_kind, from_executable, from_stat = \
653
self.source._comparison_data(from_entry, from_path)
656
from_versioned = False
660
from_executable = None
661
versioned = (from_versioned, True)
662
to_kind, to_executable, to_stat = \
663
self.target._comparison_data(to_entry, to_path)
664
kind = (from_kind, to_kind)
665
if kind[0] != kind[1]:
666
changed_content = True
667
elif from_kind == 'file':
668
from_size = self.source._file_size(from_entry, from_stat)
669
to_size = self.target._file_size(to_entry, to_stat)
670
if from_size != to_size:
671
changed_content = True
672
elif (self.source.get_file_sha1(file_id, from_path, from_stat) !=
673
self.target.get_file_sha1(file_id, to_path, to_stat)):
674
changed_content = True
675
elif from_kind == 'symlink':
676
if (self.source.get_symlink_target(file_id) !=
677
self.target.get_symlink_target(file_id)):
678
changed_content = True
679
parent = (from_parent, to_entry.parent_id)
680
name = (from_name, to_entry.name)
681
executable = (from_executable, to_executable)
683
pb.update('comparing files', entry_count, num_entries)
684
if (changed_content is not False or versioned[0] != versioned[1]
685
or parent[0] != parent[1] or name[0] != name[1] or
686
executable[0] != executable[1] or include_unchanged):
687
yield (file_id, to_path, changed_content, versioned, parent,
688
name, kind, executable)
690
def get_to_path(from_entry):
691
if from_entry.parent_id is None:
694
if from_entry.parent_id not in to_paths:
695
get_to_path(self.source.inventory[from_entry.parent_id])
696
to_path = osutils.pathjoin(to_paths[from_entry.parent_id],
698
to_paths[from_entry.file_id] = to_path
701
for path, from_entry in from_entries_by_dir:
702
file_id = from_entry.file_id
703
if file_id in to_paths:
705
to_path = get_to_path(from_entry)
708
pb.update('comparing files', entry_count, num_entries)
709
versioned = (True, False)
710
parent = (from_entry.parent_id, None)
711
name = (from_entry.name, None)
712
from_kind, from_executable, stat_value = \
713
self.source._comparison_data(from_entry, path)
714
kind = (from_kind, None)
715
executable = (from_executable, None)
716
changed_content = True
717
# the parent's path is necessarily known at this point.
718
yield(file_id, to_path, changed_content, versioned, parent,
719
name, kind, executable)
722
# This was deprecated before 0.12, but did not have an official warning
723
@symbol_versioning.deprecated_function(symbol_versioning.zero_twelve)
724
def RevisionTree(*args, **kwargs):
725
"""RevisionTree has moved to bzrlib.revisiontree.RevisionTree()
727
Accessing it as bzrlib.tree.RevisionTree has been deprecated as of
730
from bzrlib.revisiontree import RevisionTree as _RevisionTree
731
return _RevisionTree(*args, **kwargs)
added
removed
bzrlib/tree.py
