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_file_ids=None, pb=None):
92
intertree = InterTree.get(from_tree, self)
93
return intertree._iter_changes(include_unchanged,
94
specific_file_ids, pb)
97
"""Get a list of the conflicts in the tree.
99
Each conflict is an instance of bzrlib.conflicts.Conflict.
103
def get_parent_ids(self):
104
"""Get the parent ids for this tree.
106
:return: a list of parent ids. [] is returned to indicate
107
a tree with no parents.
108
:raises: BzrError if the parents are not known.
110
raise NotImplementedError(self.get_parent_ids)
112
def has_filename(self, filename):
113
"""True if the tree has given filename."""
114
raise NotImplementedError()
116
def has_id(self, file_id):
117
file_id = osutils.safe_file_id(file_id)
118
return self.inventory.has_id(file_id)
120
__contains__ = has_id
122
def has_or_had_id(self, file_id):
123
file_id = osutils.safe_file_id(file_id)
124
if file_id == self.inventory.root.file_id:
126
return self.inventory.has_id(file_id)
129
return iter(self.inventory)
131
def id2path(self, file_id):
132
file_id = osutils.safe_file_id(file_id)
133
return self.inventory.id2path(file_id)
135
def is_control_filename(self, filename):
136
"""True if filename is the name of a control file in this tree.
138
:param filename: A filename within the tree. This is a relative path
139
from the root of this tree.
141
This is true IF and ONLY IF the filename is part of the meta data
142
that bzr controls in this tree. I.E. a random .bzr directory placed
143
on disk will not be a control file for this tree.
145
return self.bzrdir.is_control_filename(filename)
148
def iter_entries_by_dir(self, specific_file_ids=None):
149
"""Walk the tree in 'by_dir' order.
151
This will yield each entry in the tree as a (path, entry) tuple. The
152
order that they are yielded is: the contents of a directory are
153
preceeded by the parent of a directory, and all the contents of a
154
directory are grouped together.
156
return self.inventory.iter_entries_by_dir(
157
specific_file_ids=specific_file_ids)
159
def kind(self, file_id):
160
raise NotImplementedError("subclasses must implement kind")
162
def _comparison_data(self, entry, path):
163
"""Return a tuple of kind, executable, stat_value for a file.
165
entry may be None if there is no inventory entry for the file, but
166
path must always be supplied.
168
kind is None if there is no file present (even if an inventory id is
169
present). executable is False for non-file entries.
171
raise NotImplementedError(self._comparison_data)
173
def _file_size(self, entry, stat_value):
174
raise NotImplementedError(self._file_size)
176
def _get_inventory(self):
177
return self._inventory
179
def get_file(self, file_id):
180
"""Return a file object for the file file_id in the tree."""
181
raise NotImplementedError(self.get_file)
183
def get_file_by_path(self, path):
184
return self.get_file(self._inventory.path2id(path))
186
def annotate_iter(self, file_id):
187
"""Return an iterator of revision_id, line tuples
189
For working trees (and mutable trees in general), the special
190
revision_id 'current:' will be used for lines that are new in this
191
tree, e.g. uncommitted changes.
192
:param file_id: The file to produce an annotated version from
194
raise NotImplementedError(self.annotate_iter)
196
inventory = property(_get_inventory,
197
doc="Inventory of this Tree")
199
def _check_retrieved(self, ie, f):
202
fp = fingerprint_file(f)
205
if ie.text_size is not None:
206
if ie.text_size != fp['size']:
207
raise BzrError("mismatched size for file %r in %r" % (ie.file_id, self._store),
208
["inventory expects %d bytes" % ie.text_size,
209
"file is actually %d bytes" % fp['size'],
210
"store is probably damaged/corrupt"])
212
if ie.text_sha1 != fp['sha1']:
213
raise BzrError("wrong SHA-1 for file %r in %r" % (ie.file_id, self._store),
214
["inventory expects %s" % ie.text_sha1,
215
"file is actually %s" % fp['sha1'],
216
"store is probably damaged/corrupt"])
218
def path2id(self, path):
219
"""Return the id for path in this tree."""
220
return self._inventory.path2id(path)
222
def paths2ids(self, paths, trees=[], require_versioned=True):
223
"""Return all the ids that can be reached by walking from paths.
225
Each path is looked up in each this tree and any extras provided in
226
trees, and this is repeated recursively: the children in an extra tree
227
of a directory that has been renamed under a provided path in this tree
228
are all returned, even if none exist until a provided path in this
229
tree, and vice versa.
231
:param paths: An iterable of paths to start converting to ids from.
232
Alternatively, if paths is None, no ids should be calculated and None
233
will be returned. This is offered to make calling the api unconditional
234
for code that *might* take a list of files.
235
:param trees: Additional trees to consider.
236
:param require_versioned: If False, do not raise NotVersionedError if
237
an element of paths is not versioned in this tree and all of trees.
239
return find_ids_across_trees(paths, [self] + list(trees), require_versioned)
241
def print_file(self, file_id):
242
"""Print file with id `file_id` to stdout."""
243
file_id = osutils.safe_file_id(file_id)
245
sys.stdout.write(self.get_file_text(file_id))
250
def revision_tree(self, revision_id):
251
"""Obtain a revision tree for the revision revision_id.
253
The intention of this method is to allow access to possibly cached
254
tree data. Implementors of this method should raise NoSuchRevision if
255
the tree is not locally available, even if they could obtain the
256
tree via a repository or some other means. Callers are responsible
257
for finding the ultimate source for a revision tree.
259
:param revision_id: The revision_id of the requested tree.
261
:raises: NoSuchRevision if the tree cannot be obtained.
263
raise errors.NoSuchRevisionInTree(self, revision_id)
266
"""What files are present in this tree and unknown.
268
:return: an iterator over the unknown files.
275
def filter_unversioned_files(self, paths):
276
"""Filter out paths that are not versioned.
278
:return: set of paths.
280
# NB: we specifically *don't* call self.has_filename, because for
281
# WorkingTrees that can indicate files that exist on disk but that
283
pred = self.inventory.has_filename
284
return set((p for p in paths if not pred(p)))
286
def walkdirs(self, prefix=""):
287
"""Walk the contents of this tree from path down.
289
This yields all the data about the contents of a directory at a time.
290
After each directory has been yielded, if the caller has mutated the
291
list to exclude some directories, they are then not descended into.
293
The data yielded is of the form:
294
((directory-relpath, directory-path-from-root, directory-fileid),
295
[(relpath, basename, kind, lstat, path_from_tree_root, file_id,
296
versioned_kind), ...]),
297
- directory-relpath is the containing dirs relpath from prefix
298
- directory-path-from-root is the containing dirs path from /
299
- directory-fileid is the id of the directory if it is versioned.
300
- relpath is the relative path within the subtree being walked.
301
- basename is the basename
302
- kind is the kind of the file now. If unknonwn then the file is not
303
present within the tree - but it may be recorded as versioned. See
305
- lstat is the stat data *if* the file was statted.
306
- path_from_tree_root is the path from the root of the tree.
307
- file_id is the file_id is the entry is versioned.
308
- versioned_kind is the kind of the file as last recorded in the
309
versioning system. If 'unknown' the file is not versioned.
310
One of 'kind' and 'versioned_kind' must not be 'unknown'.
312
:param prefix: Start walking from prefix within the tree rather than
313
at the root. This allows one to walk a subtree but get paths that are
314
relative to a tree rooted higher up.
315
:return: an iterator over the directory data.
317
raise NotImplementedError(self.walkdirs)
320
class EmptyTree(Tree):
323
self._inventory = Inventory(root_id=None)
324
symbol_versioning.warn('EmptyTree is deprecated as of bzr 0.9 please'
325
' use repository.revision_tree instead.',
326
DeprecationWarning, stacklevel=2)
328
def get_parent_ids(self):
331
def get_symlink_target(self, file_id):
334
def has_filename(self, filename):
337
def kind(self, file_id):
338
file_id = osutils.safe_file_id(file_id)
339
assert self._inventory[file_id].kind == "directory"
342
def list_files(self, include_root=False):
345
def __contains__(self, file_id):
346
file_id = osutils.safe_file_id(file_id)
347
return (file_id in self._inventory)
349
def get_file_sha1(self, file_id, path=None, stat_value=None):
353
######################################################################
356
# TODO: Merge these two functions into a single one that can operate
357
# on either a whole tree or a set of files.
359
# TODO: Return the diff in order by filename, not by category or in
360
# random order. Can probably be done by lock-stepping through the
361
# filenames from both trees.
364
def file_status(filename, old_tree, new_tree):
365
"""Return single-letter status, old and new names for a file.
367
The complexity here is in deciding how to represent renames;
368
many complex cases are possible.
370
old_inv = old_tree.inventory
371
new_inv = new_tree.inventory
372
new_id = new_inv.path2id(filename)
373
old_id = old_inv.path2id(filename)
375
if not new_id and not old_id:
376
# easy: doesn't exist in either; not versioned at all
377
if new_tree.is_ignored(filename):
378
return 'I', None, None
380
return '?', None, None
382
# There is now a file of this name, great.
385
# There is no longer a file of this name, but we can describe
386
# what happened to the file that used to have
387
# this name. There are two possibilities: either it was
388
# deleted entirely, or renamed.
390
if new_inv.has_id(old_id):
391
return 'X', old_inv.id2path(old_id), new_inv.id2path(old_id)
393
return 'D', old_inv.id2path(old_id), None
395
# if the file_id is new in this revision, it is added
396
if new_id and not old_inv.has_id(new_id):
399
# if there used to be a file of this name, but that ID has now
400
# disappeared, it is deleted
401
if old_id and not new_inv.has_id(old_id):
408
def find_renames(old_inv, new_inv):
409
for file_id in old_inv:
410
if file_id not in new_inv:
412
old_name = old_inv.id2path(file_id)
413
new_name = new_inv.id2path(file_id)
414
if old_name != new_name:
415
yield (old_name, new_name)
418
def find_ids_across_trees(filenames, trees, require_versioned=True):
419
"""Find the ids corresponding to specified filenames.
421
All matches in all trees will be used, and all children of matched
422
directories will be used.
424
:param filenames: The filenames to find file_ids for (if None, returns
426
:param trees: The trees to find file_ids within
427
:param require_versioned: if true, all specified filenames must occur in
429
:return: a set of file ids for the specified filenames and their children.
433
specified_path_ids = _find_ids_across_trees(filenames, trees,
435
return _find_children_across_trees(specified_path_ids, trees)
436
# specified_ids = [id for path, id in _find_path_ids_across_trees(filenames, trees, require_versioned)]
437
# return _find_children_across_trees(specified_ids, trees)
439
def find_path_ids_across_trees(filenames, trees, require_versioned=True):
440
"""Find the paths and ids corresponding to specified filenames.
442
All matches in all trees will be used, and all children of matched
443
directories will be included
445
:param filenames: The filenames to find file_ids for
446
:param trees: The trees to find file_ids within
447
:param require_versioned: if true, all specified filenames must occur in
449
:return: a set of (path, file ids) for the specified filenames and their
450
children. The returned path is the path of the id in the first tree
451
that contains it. This matters when files have been moved
455
# This function needs to know the ids for filenames in all trees, then
456
# search for those same files and children in all the other trees.
457
# it is complicated by the same path in two trees being able to have
458
# different ids, which might both be present in both trees.
459
# consider two trees, which have had 'mv foo bar' and 'mv baz foo' done
460
# in this case, a diff of 'foo' should should changes to both the current
461
# 'bar' and the current 'foo' which was baz. Its arguable that if
462
# the situation is 'mv parent/foo bar' and 'mv baz parent/foo', that
463
# we should return the current bar and the current parent/foo' - at the
464
# moment we do, but we loop around all ids and all trees: I*T checks.
466
# Updating this algorithm to be fast in the common case:
467
# nothing has moved, all files have the same id in parent, child and there
468
# are only two trees (or one is working tree and the others are parents).
469
# walk the dirstate. as we find each path, gather the paths of that
470
# id in all trees. add a mapping from the id to the path in those trees.
471
# now lookup children by id, again in all trees; for these trees that
472
# nothing has moved in, the id->path mapping will allow us to find the
473
# parent trivially. To answer 'has anything been moved' in one of the
474
# dirstate parent trees though, we will need to stare harder at it.
476
# Now, given a path index, that is trivial for any one tree, and given
477
# that we can ask for additional data from a dirstate tree, its a single
478
# pass, though it will require scanning the entire tree to find paths
479
# that were at the current location.
480
# ideal results?: There are three things: tree, path, id. Pathologically
481
# we can have completely disjoint ids for each tree; but we cannot have
482
# disjoin paths for each tree, except if we scan each tree for the
483
# different ids from other trees.
485
specified_path_ids = _find_ids_across_trees(filenames, trees,
487
return _find_path_id_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
"""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.
575
# NB: show_status depends on being able to pass in non-versioned files
576
# and report them as unknown
577
trees = (self.source,)
578
if extra_trees is not None:
579
trees = trees + tuple(extra_trees)
580
# target is usually the newer tree:
581
specific_file_ids = self.target.paths2ids(specific_files, trees,
582
require_versioned=require_versioned)
583
if specific_files and not specific_file_ids:
584
# All files are unversioned, so just return an empty delta
585
# _compare_trees would think we want a complete delta
586
return delta.TreeDelta()
587
return delta._compare_trees(self.source, self.target, want_unchanged,
588
specific_file_ids, include_root)
590
def _iter_changes(self, include_unchanged=False,
591
specific_file_ids=None, pb=None):
592
"""Generate an iterator of changes between trees.
595
(file_id, path, changed_content, versioned, parent, name, kind,
598
Path is relative to the target tree. changed_content is True if the
599
file's content has changed. This includes changes to its kind, and to
602
versioned, parent, name, kind, executable are tuples of (from, to).
603
If a file is missing in a tree, its kind is None.
605
Iteration is done in parent-to-child order, relative to the target
609
from_entries_by_dir = list(self.source.inventory.iter_entries_by_dir(
610
specific_file_ids=specific_file_ids))
611
from_data = dict((e.file_id, (p, e)) for p, e in from_entries_by_dir)
612
to_entries_by_dir = list(self.target.inventory.iter_entries_by_dir(
613
specific_file_ids=specific_file_ids))
614
num_entries = len(from_entries_by_dir) + len(to_entries_by_dir)
616
for to_path, to_entry in to_entries_by_dir:
617
file_id = to_entry.file_id
618
to_paths[file_id] = to_path
620
changed_content = False
621
from_path, from_entry = from_data.get(file_id, (None, None))
622
from_versioned = (from_entry is not None)
623
if from_entry is not None:
624
from_versioned = True
625
from_name = from_entry.name
626
from_parent = from_entry.parent_id
627
from_kind, from_executable, from_stat = \
628
self.source._comparison_data(from_entry, from_path)
631
from_versioned = False
635
from_executable = None
636
versioned = (from_versioned, True)
637
to_kind, to_executable, to_stat = \
638
self.target._comparison_data(to_entry, to_path)
639
kind = (from_kind, to_kind)
640
if kind[0] != kind[1]:
641
changed_content = True
642
elif from_kind == 'file':
643
from_size = self.source._file_size(from_entry, from_stat)
644
to_size = self.target._file_size(to_entry, to_stat)
645
if from_size != to_size:
646
changed_content = True
647
elif (self.source.get_file_sha1(file_id, from_path, from_stat) !=
648
self.target.get_file_sha1(file_id, to_path, to_stat)):
649
changed_content = True
650
elif from_kind == 'symlink':
651
if (self.source.get_symlink_target(file_id) !=
652
self.target.get_symlink_target(file_id)):
653
changed_content = True
654
parent = (from_parent, to_entry.parent_id)
655
name = (from_name, to_entry.name)
656
executable = (from_executable, to_executable)
658
pb.update('comparing files', entry_count, num_entries)
659
if (changed_content is not False or versioned[0] != versioned[1]
660
or parent[0] != parent[1] or name[0] != name[1] or
661
executable[0] != executable[1] or include_unchanged):
662
yield (file_id, to_path, changed_content, versioned, parent,
663
name, kind, executable)
665
def get_to_path(from_entry):
666
if from_entry.parent_id is None:
669
if from_entry.parent_id not in to_paths:
670
get_to_path(self.source.inventory[from_entry.parent_id])
671
to_path = osutils.pathjoin(to_paths[from_entry.parent_id],
673
to_paths[from_entry.file_id] = to_path
676
for path, from_entry in from_entries_by_dir:
677
file_id = from_entry.file_id
678
if file_id in to_paths:
680
to_path = get_to_path(from_entry)
683
pb.update('comparing files', entry_count, num_entries)
684
versioned = (True, False)
685
parent = (from_entry.parent_id, None)
686
name = (from_entry.name, None)
687
from_kind, from_executable, stat_value = \
688
self.source._comparison_data(from_entry, path)
689
kind = (from_kind, None)
690
executable = (from_executable, None)
691
changed_content = True
692
# the parent's path is necessarily known at this point.
693
yield(file_id, to_path, changed_content, versioned, parent,
694
name, kind, executable)
697
# This was deprecated before 0.12, but did not have an official warning
698
@symbol_versioning.deprecated_function(symbol_versioning.zero_twelve)
699
def RevisionTree(*args, **kwargs):
700
"""RevisionTree has moved to bzrlib.revisiontree.RevisionTree()
702
Accessing it as bzrlib.tree.RevisionTree has been deprecated as of
705
from bzrlib.revisiontree import RevisionTree as _RevisionTree
706
return _RevisionTree(*args, **kwargs)
added
removed
bzrlib/tree.py
