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(from_tree, self, 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
return self.inventory.has_id(file_id)
119
__contains__ = has_id
121
def has_or_had_id(self, file_id):
122
if file_id == self.inventory.root.file_id:
124
return self.inventory.has_id(file_id)
127
return iter(self.inventory)
129
def id2path(self, file_id):
130
return self.inventory.id2path(file_id)
132
def is_control_filename(self, filename):
133
"""True if filename is the name of a control file in this tree.
135
:param filename: A filename within the tree. This is a relative path
136
from the root of this tree.
138
This is true IF and ONLY IF the filename is part of the meta data
139
that bzr controls in this tree. I.E. a random .bzr directory placed
140
on disk will not be a control file for this tree.
142
return self.bzrdir.is_control_filename(filename)
145
def iter_entries_by_dir(self, specific_file_ids=None):
146
"""Walk the tree in 'by_dir' order.
148
This will yield each entry in the tree as a (path, entry) tuple. The
149
order that they are yielded is: the contents of a directory are
150
preceeded by the parent of a directory, and all the contents of a
151
directory are grouped together.
153
return self.inventory.iter_entries_by_dir(
154
specific_file_ids=specific_file_ids)
156
def kind(self, file_id):
157
raise NotImplementedError("subclasses must implement kind")
159
def _comparison_data(self, entry, path):
160
"""Return a tuple of kind, executable, stat_value for a file.
162
entry may be None if there is no inventory entry for the file, but
163
path must always be supplied.
165
kind is None if there is no file present (even if an inventory id is
166
present). executable is False for non-file entries.
168
raise NotImplementedError(self._comparison_data)
170
def _file_size(self, entry, stat_value):
171
raise NotImplementedError(self._file_size)
173
def _get_inventory(self):
174
return self._inventory
176
def get_file(self, file_id):
177
"""Return a file object for the file file_id in the tree."""
178
raise NotImplementedError(self.get_file)
180
def get_file_by_path(self, path):
181
return self.get_file(self._inventory.path2id(path))
183
def annotate_iter(self, file_id):
184
"""Return an iterator of revision_id, line tuples
186
For working trees (and mutable trees in general), the special
187
revision_id 'current:' will be used for lines that are new in this
188
tree, e.g. uncommitted changes.
189
:param file_id: The file to produce an annotated version from
191
raise NotImplementedError(self.annotate_iter)
193
inventory = property(_get_inventory,
194
doc="Inventory of this Tree")
196
def _check_retrieved(self, ie, f):
199
fp = fingerprint_file(f)
202
if ie.text_size is not None:
203
if ie.text_size != fp['size']:
204
raise BzrError("mismatched size for file %r in %r" % (ie.file_id, self._store),
205
["inventory expects %d bytes" % ie.text_size,
206
"file is actually %d bytes" % fp['size'],
207
"store is probably damaged/corrupt"])
209
if ie.text_sha1 != fp['sha1']:
210
raise BzrError("wrong SHA-1 for file %r in %r" % (ie.file_id, self._store),
211
["inventory expects %s" % ie.text_sha1,
212
"file is actually %s" % fp['sha1'],
213
"store is probably damaged/corrupt"])
215
def path2id(self, path):
216
"""Return the id for path in this tree."""
217
return self._inventory.path2id(path)
219
def print_file(self, file_id):
220
"""Print file with id `file_id` to stdout."""
222
sys.stdout.write(self.get_file_text(file_id))
227
def revision_tree(self, revision_id):
228
"""Obtain a revision tree for the revision revision_id.
230
The intention of this method is to allow access to possibly cached
231
tree data. Implementors of this method should raise NoSuchRevision if
232
the tree is not locally available, even if they could obtain the
233
tree via a repository or some other means. Callers are responsible
234
for finding the ultimate source for a revision tree.
236
:param revision_id: The revision_id of the requested tree.
238
:raises: NoSuchRevision if the tree cannot be obtained.
240
raise errors.NoSuchRevisionInTree(self, revision_id)
243
"""What files are present in this tree and unknown.
245
:return: an iterator over the unknown files.
252
def filter_unversioned_files(self, paths):
253
"""Filter out paths that are not versioned.
255
:return: set of paths.
257
# NB: we specifically *don't* call self.has_filename, because for
258
# WorkingTrees that can indicate files that exist on disk but that
260
pred = self.inventory.has_filename
261
return set((p for p in paths if not pred(p)))
263
def walkdirs(self, prefix=""):
264
"""Walk the contents of this tree from path down.
266
This yields all the data about the contents of a directory at a time.
267
After each directory has been yielded, if the caller has mutated the
268
list to exclude some directories, they are then not descended into.
270
The data yielded is of the form:
271
((directory-relpath, directory-path-from-root, directory-fileid),
272
[(relpath, basename, kind, lstat, path_from_tree_root, file_id,
273
versioned_kind), ...]),
274
- directory-relpath is the containing dirs relpath from prefix
275
- directory-path-from-root is the containing dirs path from /
276
- directory-fileid is the id of the directory if it is versioned.
277
- relpath is the relative path within the subtree being walked.
278
- basename is the basename
279
- kind is the kind of the file now. If unknonwn then the file is not
280
present within the tree - but it may be recorded as versioned. See
282
- lstat is the stat data *if* the file was statted.
283
- path_from_tree_root is the path from the root of the tree.
284
- file_id is the file_id is the entry is versioned.
285
- versioned_kind is the kind of the file as last recorded in the
286
versioning system. If 'unknown' the file is not versioned.
287
One of 'kind' and 'versioned_kind' must not be 'unknown'.
289
:param prefix: Start walking from prefix within the tree rather than
290
at the root. This allows one to walk a subtree but get paths that are
291
relative to a tree rooted higher up.
292
:return: an iterator over the directory data.
294
raise NotImplementedError(self.walkdirs)
297
class EmptyTree(Tree):
300
self._inventory = Inventory(root_id=None)
301
symbol_versioning.warn('EmptyTree is deprecated as of bzr 0.9 please'
302
' use repository.revision_tree instead.',
303
DeprecationWarning, stacklevel=2)
305
def get_parent_ids(self):
308
def get_symlink_target(self, file_id):
311
def has_filename(self, filename):
314
def kind(self, file_id):
315
assert self._inventory[file_id].kind == "directory"
318
def list_files(self, include_root=False):
321
def __contains__(self, file_id):
322
return (file_id in self._inventory)
324
def get_file_sha1(self, file_id, path=None, stat_value=None):
328
######################################################################
331
# TODO: Merge these two functions into a single one that can operate
332
# on either a whole tree or a set of files.
334
# TODO: Return the diff in order by filename, not by category or in
335
# random order. Can probably be done by lock-stepping through the
336
# filenames from both trees.
339
def file_status(filename, old_tree, new_tree):
340
"""Return single-letter status, old and new names for a file.
342
The complexity here is in deciding how to represent renames;
343
many complex cases are possible.
345
old_inv = old_tree.inventory
346
new_inv = new_tree.inventory
347
new_id = new_inv.path2id(filename)
348
old_id = old_inv.path2id(filename)
350
if not new_id and not old_id:
351
# easy: doesn't exist in either; not versioned at all
352
if new_tree.is_ignored(filename):
353
return 'I', None, None
355
return '?', None, None
357
# There is now a file of this name, great.
360
# There is no longer a file of this name, but we can describe
361
# what happened to the file that used to have
362
# this name. There are two possibilities: either it was
363
# deleted entirely, or renamed.
365
if new_inv.has_id(old_id):
366
return 'X', old_inv.id2path(old_id), new_inv.id2path(old_id)
368
return 'D', old_inv.id2path(old_id), None
370
# if the file_id is new in this revision, it is added
371
if new_id and not old_inv.has_id(new_id):
374
# if there used to be a file of this name, but that ID has now
375
# disappeared, it is deleted
376
if old_id and not new_inv.has_id(old_id):
383
def find_renames(old_inv, new_inv):
384
for file_id in old_inv:
385
if file_id not in new_inv:
387
old_name = old_inv.id2path(file_id)
388
new_name = new_inv.id2path(file_id)
389
if old_name != new_name:
390
yield (old_name, new_name)
393
def find_ids_across_trees(filenames, trees, require_versioned=True):
394
"""Find the ids corresponding to specified filenames.
396
All matches in all trees will be used, and all children of matched
397
directories will be used.
399
:param filenames: The filenames to find file_ids for (if None, returns
401
:param trees: The trees to find file_ids within
402
:param require_versioned: if true, all specified filenames must occur in
404
:return: a set of file ids for the specified filenames and their children.
408
specified_path_ids = _find_ids_across_trees(filenames, trees,
410
return _find_children_across_trees(specified_path_ids, trees)
411
# specified_ids = [id for path, id in _find_path_ids_across_trees(filenames, trees, require_versioned)]
412
# return _find_children_across_trees(specified_ids, trees)
414
def find_path_ids_across_trees(filenames, trees, require_versioned=True):
415
"""Find the paths and ids corresponding to specified filenames.
417
All matches in all trees will be used, and all children of matched
418
directories will be included
420
:param filenames: The filenames to find file_ids for
421
:param trees: The trees to find file_ids within
422
:param require_versioned: if true, all specified filenames must occur in
424
:return: a set of (path, file ids) for the specified filenames and their
425
children. The returned path is the path of the id in the first tree
426
that contains it. This matters when files have been moved
430
# This function needs to know the ids for filenames in all trees, then
431
# search for those same files and children in all the other trees.
432
# it is complicated by the same path in two trees being able to have
433
# different ids, which might both be present in both trees.
434
# consider two trees, which have had 'mv foo bar' and 'mv baz foo' done
435
# in this case, a diff of 'foo' should should changes to both the current
436
# 'bar' and the current 'foo' which was baz. Its arguable that if
437
# the situation is 'mv parent/foo bar' and 'mv baz parent/foo', that
438
# we should return the current bar and the current parent/foo' - at the
439
# moment we do, but we loop around all ids and all trees: I*T checks.
441
# Updating this algorithm to be fast in the common case:
442
# nothing has moved, all files have the same id in parent, child and there
443
# are only two trees (or one is working tree and the others are parents).
444
# walk the dirstate. as we find each path, gather the paths of that
445
# id in all trees. add a mapping from the id to the path in those trees.
446
# now lookup children by id, again in all trees; for these trees that
447
# nothing has moved in, the id->path mapping will allow us to find the
448
# parent trivially. To answer 'has anything been moved' in one of the
449
# dirstate parent trees though, we will need to stare harder at it.
451
# Now, given a path index, that is trivial for any one tree, and given
452
# that we can ask for additional data from a dirstate tree, its a single
453
# pass, though it will require scanning the entire tree to find paths
454
# that were at the current location.
455
# ideal results?: There are three things: tree, path, id. Pathologically
456
# we can have completely disjoint ids for each tree; but we cannot have
457
# disjoin paths for each tree, except if we scan each tree for the
458
# different ids from other trees.
460
specified_path_ids = _find_ids_across_trees(filenames, trees,
462
return _find_path_id_children_across_trees(specified_path_ids, trees)
465
def _find_ids_across_trees(filenames, trees, require_versioned):
466
"""Find the ids corresponding to specified filenames.
468
All matches in all trees will be used, but subdirectories are not scanned.
470
:param filenames: The filenames to find file_ids for
471
:param trees: The trees to find file_ids within
472
:param require_versioned: if true, all specified filenames must occur in
474
:return: a set of (path, file ids) for the specified filenames
477
interesting_ids = set()
478
for tree_path in filenames:
481
file_id = tree.path2id(tree_path)
482
if file_id is not None:
483
interesting_ids.add(file_id)
486
not_versioned.append(tree_path)
487
if len(not_versioned) > 0 and require_versioned:
488
raise errors.PathsNotVersionedError(not_versioned)
489
return interesting_ids
492
def _find_children_across_trees(specified_ids, trees):
493
"""Return a set including specified ids and their children
495
All matches in all trees will be used.
497
:param trees: The trees to find file_ids within
498
:return: a set containing all specified ids and their children
500
interesting_ids = set(specified_ids)
501
pending = interesting_ids
502
# now handle children of interesting ids
503
# we loop so that we handle all children of each id in both trees
504
while len(pending) > 0:
506
for file_id in pending:
508
if not tree.has_id(file_id):
510
entry = tree.inventory[file_id]
511
for child in getattr(entry, 'children', {}).itervalues():
512
if child.file_id not in interesting_ids:
513
new_pending.add(child.file_id)
514
interesting_ids.update(new_pending)
515
pending = new_pending
516
return interesting_ids
519
class InterTree(InterObject):
520
"""This class represents operations taking place between two Trees.
522
Its instances have methods like 'compare' and contain references to the
523
source and target trees these operations are to be carried out on.
525
clients of bzrlib should not need to use InterTree directly, rather they
526
should use the convenience methods on Tree such as 'Tree.compare()' which
527
will pass through to InterTree as appropriate.
533
def compare(self, want_unchanged=False, specific_files=None,
534
extra_trees=None, require_versioned=False, include_root=False):
535
"""Return the changes from source to target.
537
:return: A TreeDelta.
538
:param specific_files: An optional list of file paths to restrict the
539
comparison to. When mapping filenames to ids, all matches in all
540
trees (including optional extra_trees) are used, and all children of
541
matched directories are included.
542
:param want_unchanged: An optional boolean requesting the inclusion of
543
unchanged entries in the result.
544
:param extra_trees: An optional list of additional trees to use when
545
mapping the contents of specific_files (paths) to file_ids.
546
:param require_versioned: An optional boolean (defaults to False). When
547
supplied and True all the 'specific_files' must be versioned, or
548
a PathsNotVersionedError will be thrown.
550
# NB: show_status depends on being able to pass in non-versioned files
551
# and report them as unknown
552
trees = (self.source, self.target)
553
if extra_trees is not None:
554
trees = trees + tuple(extra_trees)
555
specific_file_ids = find_ids_across_trees(specific_files,
556
trees, require_versioned=require_versioned)
557
if specific_files and not specific_file_ids:
558
# All files are unversioned, so just return an empty delta
559
# _compare_trees would think we want a complete delta
560
return delta.TreeDelta()
561
return delta._compare_trees(self.source, self.target, want_unchanged,
562
specific_file_ids, include_root)
564
def _iter_changes(self, from_tree, to_tree, include_unchanged,
565
specific_file_ids, pb):
566
"""Generate an iterator of changes between trees.
569
(file_id, path, changed_content, versioned, parent, name, kind,
572
Path is relative to the to_tree. changed_content is True if the file's
573
content has changed. This includes changes to its kind, and to
576
versioned, parent, name, kind, executable are tuples of (from, to).
577
If a file is missing in a tree, its kind is None.
579
Iteration is done in parent-to-child order, relative to the to_tree.
582
from_entries_by_dir = list(from_tree.inventory.iter_entries_by_dir(
583
specific_file_ids=specific_file_ids))
584
from_data = dict((e.file_id, (p, e)) for p, e in from_entries_by_dir)
585
to_entries_by_dir = list(to_tree.inventory.iter_entries_by_dir(
586
specific_file_ids=specific_file_ids))
587
num_entries = len(from_entries_by_dir) + len(to_entries_by_dir)
589
for to_path, to_entry in to_entries_by_dir:
590
file_id = to_entry.file_id
591
to_paths[file_id] = to_path
593
changed_content = False
594
from_path, from_entry = from_data.get(file_id, (None, None))
595
from_versioned = (from_entry is not None)
596
if from_entry is not None:
597
from_versioned = True
598
from_name = from_entry.name
599
from_parent = from_entry.parent_id
600
from_kind, from_executable, from_stat = \
601
from_tree._comparison_data(from_entry, from_path)
604
from_versioned = False
608
from_executable = None
609
versioned = (from_versioned, True)
610
to_kind, to_executable, to_stat = \
611
to_tree._comparison_data(to_entry, to_path)
612
kind = (from_kind, to_kind)
613
if kind[0] != kind[1]:
614
changed_content = True
615
elif from_kind == 'file':
616
from_size = from_tree._file_size(from_entry, from_stat)
617
to_size = to_tree._file_size(to_entry, to_stat)
618
if from_size != to_size:
619
changed_content = True
620
elif (from_tree.get_file_sha1(file_id, from_path, from_stat) !=
621
to_tree.get_file_sha1(file_id, to_path, to_stat)):
622
changed_content = True
623
elif from_kind == 'symlink':
624
if (from_tree.get_symlink_target(file_id) !=
625
to_tree.get_symlink_target(file_id)):
626
changed_content = True
627
parent = (from_parent, to_entry.parent_id)
628
name = (from_name, to_entry.name)
629
executable = (from_executable, to_executable)
631
pb.update('comparing files', entry_count, num_entries)
632
if (changed_content is not False or versioned[0] != versioned[1]
633
or parent[0] != parent[1] or name[0] != name[1] or
634
executable[0] != executable[1] or include_unchanged):
635
yield (file_id, to_path, changed_content, versioned, parent,
636
name, kind, executable)
638
def get_to_path(from_entry):
639
if from_entry.parent_id is None:
642
if from_entry.parent_id not in to_paths:
643
get_to_path(from_tree.inventory[from_entry.parent_id])
644
to_path = osutils.pathjoin(to_paths[from_entry.parent_id],
646
to_paths[from_entry.file_id] = to_path
649
for path, from_entry in from_entries_by_dir:
650
file_id = from_entry.file_id
651
if file_id in to_paths:
653
to_path = get_to_path(from_entry)
656
pb.update('comparing files', entry_count, num_entries)
657
versioned = (True, False)
658
parent = (from_entry.parent_id, None)
659
name = (from_entry.name, None)
660
from_kind, from_executable, stat_value = \
661
from_tree._comparison_data(from_entry, path)
662
kind = (from_kind, None)
663
executable = (from_executable, None)
664
changed_content = True
665
# the parent's path is necessarily known at this point.
666
yield(file_id, to_path, changed_content, versioned, parent,
667
name, kind, executable)
670
# This was deprecated before 0.12, but did not have an official warning
671
@symbol_versioning.deprecated_function(symbol_versioning.zero_twelve)
672
def RevisionTree(*args, **kwargs):
673
"""RevisionTree has moved to bzrlib.revisiontree.RevisionTree()
675
Accessing it as bzrlib.tree.RevisionTree has been deprecated as of
678
from bzrlib.revisiontree import RevisionTree as _RevisionTree
679
return _RevisionTree(*args, **kwargs)
added
removed
bzrlib/tree.py
