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)
138
return iter(self.inventory)
140
def id2path(self, file_id):
141
file_id = osutils.safe_file_id(file_id)
142
return self.inventory.id2path(file_id)
144
def is_control_filename(self, filename):
145
"""True if filename is the name of a control file in this tree.
147
:param filename: A filename within the tree. This is a relative path
148
from the root of this tree.
150
This is true IF and ONLY IF the filename is part of the meta data
151
that bzr controls in this tree. I.E. a random .bzr directory placed
152
on disk will not be a control file for this tree.
154
return self.bzrdir.is_control_filename(filename)
157
def iter_entries_by_dir(self, specific_file_ids=None):
158
"""Walk the tree in 'by_dir' order.
160
This will yield each entry in the tree as a (path, entry) tuple. The
161
order that they are yielded is: the contents of a directory are
162
preceeded by the parent of a directory, and all the contents of a
163
directory are grouped together.
165
return self.inventory.iter_entries_by_dir(
166
specific_file_ids=specific_file_ids)
168
def kind(self, file_id):
169
raise NotImplementedError("subclasses must implement kind")
171
def _comparison_data(self, entry, path):
172
"""Return a tuple of kind, executable, stat_value for a file.
174
entry may be None if there is no inventory entry for the file, but
175
path must always be supplied.
177
kind is None if there is no file present (even if an inventory id is
178
present). executable is False for non-file entries.
180
raise NotImplementedError(self._comparison_data)
182
def _file_size(self, entry, stat_value):
183
raise NotImplementedError(self._file_size)
185
def _get_inventory(self):
186
return self._inventory
188
def get_file(self, file_id):
189
"""Return a file object for the file file_id in the tree."""
190
raise NotImplementedError(self.get_file)
192
def get_file_mtime(self, file_id, path=None):
193
"""Return the modification time for a file.
195
:param file_id: The handle for this file.
196
:param path: The path that this file can be found at.
197
These must point to the same object.
199
raise NotImplementedError(self.get_file_mtime)
201
def get_file_by_path(self, path):
202
return self.get_file(self._inventory.path2id(path))
204
def get_symlink_target(self, file_id):
205
"""Get the target for a given file_id.
207
It is assumed that the caller already knows that file_id is referencing
209
:param file_id: Handle for the symlink entry.
210
:return: The path the symlink points to.
212
raise NotImplementedError(self.get_symlink_target)
214
def annotate_iter(self, file_id):
215
"""Return an iterator of revision_id, line tuples
217
For working trees (and mutable trees in general), the special
218
revision_id 'current:' will be used for lines that are new in this
219
tree, e.g. uncommitted changes.
220
:param file_id: The file to produce an annotated version from
222
raise NotImplementedError(self.annotate_iter)
224
inventory = property(_get_inventory,
225
doc="Inventory of this Tree")
227
def _check_retrieved(self, ie, f):
230
fp = fingerprint_file(f)
233
if ie.text_size is not None:
234
if ie.text_size != fp['size']:
235
raise BzrError("mismatched size for file %r in %r" % (ie.file_id, self._store),
236
["inventory expects %d bytes" % ie.text_size,
237
"file is actually %d bytes" % fp['size'],
238
"store is probably damaged/corrupt"])
240
if ie.text_sha1 != fp['sha1']:
241
raise BzrError("wrong SHA-1 for file %r in %r" % (ie.file_id, self._store),
242
["inventory expects %s" % ie.text_sha1,
243
"file is actually %s" % fp['sha1'],
244
"store is probably damaged/corrupt"])
246
def path2id(self, path):
247
"""Return the id for path in this tree."""
248
return self._inventory.path2id(path)
250
def paths2ids(self, paths, trees=[], require_versioned=True):
251
"""Return all the ids that can be reached by walking from paths.
253
Each path is looked up in each this tree and any extras provided in
254
trees, and this is repeated recursively: the children in an extra tree
255
of a directory that has been renamed under a provided path in this tree
256
are all returned, even if none exist until a provided path in this
257
tree, and vice versa.
259
:param paths: An iterable of paths to start converting to ids from.
260
Alternatively, if paths is None, no ids should be calculated and None
261
will be returned. This is offered to make calling the api unconditional
262
for code that *might* take a list of files.
263
:param trees: Additional trees to consider.
264
:param require_versioned: If False, do not raise NotVersionedError if
265
an element of paths is not versioned in this tree and all of trees.
267
return find_ids_across_trees(paths, [self] + list(trees), require_versioned)
269
def print_file(self, file_id):
270
"""Print file with id `file_id` to stdout."""
271
file_id = osutils.safe_file_id(file_id)
273
sys.stdout.write(self.get_file_text(file_id))
278
def revision_tree(self, revision_id):
279
"""Obtain a revision tree for the revision revision_id.
281
The intention of this method is to allow access to possibly cached
282
tree data. Implementors of this method should raise NoSuchRevision if
283
the tree is not locally available, even if they could obtain the
284
tree via a repository or some other means. Callers are responsible
285
for finding the ultimate source for a revision tree.
287
:param revision_id: The revision_id of the requested tree.
289
:raises: NoSuchRevision if the tree cannot be obtained.
291
raise errors.NoSuchRevisionInTree(self, revision_id)
294
"""What files are present in this tree and unknown.
296
:return: an iterator over the unknown files.
303
def filter_unversioned_files(self, paths):
304
"""Filter out paths that are versioned.
306
:return: set of paths.
308
# NB: we specifically *don't* call self.has_filename, because for
309
# WorkingTrees that can indicate files that exist on disk but that
311
pred = self.inventory.has_filename
312
return set((p for p in paths if not pred(p)))
314
def walkdirs(self, prefix=""):
315
"""Walk the contents of this tree from path down.
317
This yields all the data about the contents of a directory at a time.
318
After each directory has been yielded, if the caller has mutated the
319
list to exclude some directories, they are then not descended into.
321
The data yielded is of the form:
322
((directory-relpath, directory-path-from-root, directory-fileid),
323
[(relpath, basename, kind, lstat, path_from_tree_root, file_id,
324
versioned_kind), ...]),
325
- directory-relpath is the containing dirs relpath from prefix
326
- directory-path-from-root is the containing dirs path from /
327
- directory-fileid is the id of the directory if it is versioned.
328
- relpath is the relative path within the subtree being walked.
329
- basename is the basename
330
- kind is the kind of the file now. If unknonwn then the file is not
331
present within the tree - but it may be recorded as versioned. See
333
- lstat is the stat data *if* the file was statted.
334
- path_from_tree_root is the path from the root of the tree.
335
- file_id is the file_id is the entry is versioned.
336
- versioned_kind is the kind of the file as last recorded in the
337
versioning system. If 'unknown' the file is not versioned.
338
One of 'kind' and 'versioned_kind' must not be 'unknown'.
340
:param prefix: Start walking from prefix within the tree rather than
341
at the root. This allows one to walk a subtree but get paths that are
342
relative to a tree rooted higher up.
343
:return: an iterator over the directory data.
345
raise NotImplementedError(self.walkdirs)
348
class EmptyTree(Tree):
351
self._inventory = Inventory(root_id=None)
352
symbol_versioning.warn('EmptyTree is deprecated as of bzr 0.9 please'
353
' use repository.revision_tree instead.',
354
DeprecationWarning, stacklevel=2)
356
def get_parent_ids(self):
359
def get_symlink_target(self, file_id):
362
def has_filename(self, filename):
365
def kind(self, file_id):
366
file_id = osutils.safe_file_id(file_id)
367
assert self._inventory[file_id].kind == "directory"
370
def list_files(self, include_root=False):
373
def __contains__(self, file_id):
374
file_id = osutils.safe_file_id(file_id)
375
return (file_id in self._inventory)
377
def get_file_sha1(self, file_id, path=None, stat_value=None):
381
######################################################################
384
# TODO: Merge these two functions into a single one that can operate
385
# on either a whole tree or a set of files.
387
# TODO: Return the diff in order by filename, not by category or in
388
# random order. Can probably be done by lock-stepping through the
389
# filenames from both trees.
392
def file_status(filename, old_tree, new_tree):
393
"""Return single-letter status, old and new names for a file.
395
The complexity here is in deciding how to represent renames;
396
many complex cases are possible.
398
old_inv = old_tree.inventory
399
new_inv = new_tree.inventory
400
new_id = new_inv.path2id(filename)
401
old_id = old_inv.path2id(filename)
403
if not new_id and not old_id:
404
# easy: doesn't exist in either; not versioned at all
405
if new_tree.is_ignored(filename):
406
return 'I', None, None
408
return '?', None, None
410
# There is now a file of this name, great.
413
# There is no longer a file of this name, but we can describe
414
# what happened to the file that used to have
415
# this name. There are two possibilities: either it was
416
# deleted entirely, or renamed.
418
if new_inv.has_id(old_id):
419
return 'X', old_inv.id2path(old_id), new_inv.id2path(old_id)
421
return 'D', old_inv.id2path(old_id), None
423
# if the file_id is new in this revision, it is added
424
if new_id and not old_inv.has_id(new_id):
427
# if there used to be a file of this name, but that ID has now
428
# disappeared, it is deleted
429
if old_id and not new_inv.has_id(old_id):
436
def find_renames(old_inv, new_inv):
437
for file_id in old_inv:
438
if file_id not in new_inv:
440
old_name = old_inv.id2path(file_id)
441
new_name = new_inv.id2path(file_id)
442
if old_name != new_name:
443
yield (old_name, new_name)
446
def find_ids_across_trees(filenames, trees, require_versioned=True):
447
"""Find the ids corresponding to specified filenames.
449
All matches in all trees will be used, and all children of matched
450
directories will be used.
452
:param filenames: The filenames to find file_ids for (if None, returns
454
:param trees: The trees to find file_ids within
455
:param require_versioned: if true, all specified filenames must occur in
457
:return: a set of file ids for the specified filenames and their children.
461
specified_path_ids = _find_ids_across_trees(filenames, trees,
463
return _find_children_across_trees(specified_path_ids, trees)
466
def _find_ids_across_trees(filenames, trees, require_versioned):
467
"""Find the ids corresponding to specified filenames.
469
All matches in all trees will be used, but subdirectories are not scanned.
471
:param filenames: The filenames to find file_ids for
472
:param trees: The trees to find file_ids within
473
:param require_versioned: if true, all specified filenames must occur in
475
:return: a set of (path, file ids) for the specified filenames
478
interesting_ids = set()
479
for tree_path in filenames:
482
file_id = tree.path2id(tree_path)
483
if file_id is not None:
484
interesting_ids.add(file_id)
487
not_versioned.append(tree_path)
488
if len(not_versioned) > 0 and require_versioned:
489
raise errors.PathsNotVersionedError(not_versioned)
490
return interesting_ids
493
def _find_children_across_trees(specified_ids, trees):
494
"""Return a set including specified ids and their children
496
All matches in all trees will be used.
498
:param trees: The trees to find file_ids within
499
:return: a set containing all specified ids and their children
501
interesting_ids = set(specified_ids)
502
pending = interesting_ids
503
# now handle children of interesting ids
504
# we loop so that we handle all children of each id in both trees
505
while len(pending) > 0:
507
for file_id in pending:
509
if not tree.has_id(file_id):
511
entry = tree.inventory[file_id]
512
for child in getattr(entry, 'children', {}).itervalues():
513
if child.file_id not in interesting_ids:
514
new_pending.add(child.file_id)
515
interesting_ids.update(new_pending)
516
pending = new_pending
517
return interesting_ids
520
class InterTree(InterObject):
521
"""This class represents operations taking place between two Trees.
523
Its instances have methods like 'compare' and contain references to the
524
source and target trees these operations are to be carried out on.
526
clients of bzrlib should not need to use InterTree directly, rather they
527
should use the convenience methods on Tree such as 'Tree.compare()' which
528
will pass through to InterTree as appropriate.
534
def compare(self, want_unchanged=False, specific_files=None,
535
extra_trees=None, require_versioned=False, include_root=False,
536
want_unversioned=False):
537
"""Return the changes from source to target.
539
:return: A TreeDelta.
540
:param specific_files: An optional list of file paths to restrict the
541
comparison to. When mapping filenames to ids, all matches in all
542
trees (including optional extra_trees) are used, and all children of
543
matched directories are included.
544
:param want_unchanged: An optional boolean requesting the inclusion of
545
unchanged entries in the result.
546
:param extra_trees: An optional list of additional trees to use when
547
mapping the contents of specific_files (paths) to file_ids.
548
:param require_versioned: An optional boolean (defaults to False). When
549
supplied and True all the 'specific_files' must be versioned, or
550
a PathsNotVersionedError will be thrown.
551
:param want_unversioned: Scan for unversioned paths.
553
# NB: show_status depends on being able to pass in non-versioned files
554
# and report them as unknown
555
trees = (self.source,)
556
if extra_trees is not None:
557
trees = trees + tuple(extra_trees)
558
# target is usually the newer tree:
559
specific_file_ids = self.target.paths2ids(specific_files, trees,
560
require_versioned=require_versioned)
561
if specific_files and not specific_file_ids:
562
# All files are unversioned, so just return an empty delta
563
# _compare_trees would think we want a complete delta
564
result = delta.TreeDelta()
565
fake_entry = InventoryFile('unused', 'unused', 'unused')
566
result.unversioned = [(path, None,
567
self.target._comparison_data(fake_entry, path)[0]) for path in
570
return delta._compare_trees(self.source, self.target, want_unchanged,
571
specific_files, include_root, extra_trees=extra_trees,
572
want_unversioned=want_unversioned)
574
def _iter_changes(self, include_unchanged=False,
575
specific_files=None, pb=None, extra_trees=[],
576
require_versioned=True, want_unversioned=False):
577
"""Generate an iterator of changes between trees.
580
(file_id, path, changed_content, versioned, parent, name, kind,
583
Path is relative to the target tree. changed_content is True if the
584
file's content has changed. This includes changes to its kind, and to
587
versioned, parent, name, kind, executable are tuples of (from, to).
588
If a file is missing in a tree, its kind is None.
590
Iteration is done in parent-to-child order, relative to the target
593
There is no guarantee that all paths are in sorted order: the
594
requirement to expand the search due to renames may result in children
595
that should be found early being found late in the search, after
596
lexically later results have been returned.
597
:param require_versioned: Raise errors.PathsNotVersionedError if a
598
path in the specific_files list is not versioned in one of
599
source, target or extra_trees.
600
:param want_unversioned: Should unversioned files be returned in the
601
output. An unversioned file is defined as one with (False, False)
602
for the versioned pair.
604
lookup_trees = [self.source]
606
lookup_trees.extend(extra_trees)
607
specific_file_ids = self.target.paths2ids(specific_files,
608
lookup_trees, require_versioned=require_versioned)
610
all_unversioned = sorted([(p.split('/'), p) for p in self.target.extras()
611
if not specific_files or
612
osutils.is_inside_any(specific_files, p)])
613
all_unversioned = deque(all_unversioned)
615
all_unversioned = deque()
617
from_entries_by_dir = list(self.source.inventory.iter_entries_by_dir(
618
specific_file_ids=specific_file_ids))
619
from_data = dict((e.file_id, (p, e)) for p, e in from_entries_by_dir)
620
to_entries_by_dir = list(self.target.inventory.iter_entries_by_dir(
621
specific_file_ids=specific_file_ids))
622
num_entries = len(from_entries_by_dir) + len(to_entries_by_dir)
624
# the unversioned path lookup only occurs on real trees - where there
625
# can be extras. So the fake_entry is solely used to look up
626
# executable it values when execute is not supported.
627
fake_entry = InventoryFile('unused', 'unused', 'unused')
628
for to_path, to_entry in to_entries_by_dir:
629
while all_unversioned and all_unversioned[0][0] < to_path.split('/'):
630
unversioned_path = all_unversioned.popleft()
631
to_kind, to_executable, to_stat = \
632
self.target._comparison_data(fake_entry, unversioned_path[1])
633
yield (None, unversioned_path[1], True, (False, False),
635
(None, unversioned_path[0][-1]),
637
(None, to_executable))
638
file_id = to_entry.file_id
639
to_paths[file_id] = to_path
641
changed_content = False
642
from_path, from_entry = from_data.get(file_id, (None, None))
643
from_versioned = (from_entry is not None)
644
if from_entry is not None:
645
from_versioned = True
646
from_name = from_entry.name
647
from_parent = from_entry.parent_id
648
from_kind, from_executable, from_stat = \
649
self.source._comparison_data(from_entry, from_path)
652
from_versioned = False
656
from_executable = None
657
versioned = (from_versioned, True)
658
to_kind, to_executable, to_stat = \
659
self.target._comparison_data(to_entry, to_path)
660
kind = (from_kind, to_kind)
661
if kind[0] != kind[1]:
662
changed_content = True
663
elif from_kind == 'file':
664
from_size = self.source._file_size(from_entry, from_stat)
665
to_size = self.target._file_size(to_entry, to_stat)
666
if from_size != to_size:
667
changed_content = True
668
elif (self.source.get_file_sha1(file_id, from_path, from_stat) !=
669
self.target.get_file_sha1(file_id, to_path, to_stat)):
670
changed_content = True
671
elif from_kind == 'symlink':
672
if (self.source.get_symlink_target(file_id) !=
673
self.target.get_symlink_target(file_id)):
674
changed_content = True
675
parent = (from_parent, to_entry.parent_id)
676
name = (from_name, to_entry.name)
677
executable = (from_executable, to_executable)
679
pb.update('comparing files', entry_count, num_entries)
680
if (changed_content is not False or versioned[0] != versioned[1]
681
or parent[0] != parent[1] or name[0] != name[1] or
682
executable[0] != executable[1] or include_unchanged):
683
yield (file_id, to_path, changed_content, versioned, parent,
684
name, kind, executable)
685
while all_unversioned:
686
# yield any trailing unversioned paths
687
unversioned_path = all_unversioned.popleft()
688
to_kind, to_executable, to_stat = \
689
self.target._comparison_data(fake_entry, unversioned_path[1])
690
yield (None, unversioned_path[1], True, (False, False),
692
(None, unversioned_path[0][-1]),
694
(None, to_executable))
696
def get_to_path(from_entry):
697
if from_entry.parent_id is None:
700
if from_entry.parent_id not in to_paths:
701
get_to_path(self.source.inventory[from_entry.parent_id])
702
to_path = osutils.pathjoin(to_paths[from_entry.parent_id],
704
to_paths[from_entry.file_id] = to_path
707
for path, from_entry in from_entries_by_dir:
708
file_id = from_entry.file_id
709
if file_id in to_paths:
711
to_path = get_to_path(from_entry)
714
pb.update('comparing files', entry_count, num_entries)
715
versioned = (True, False)
716
parent = (from_entry.parent_id, None)
717
name = (from_entry.name, None)
718
from_kind, from_executable, stat_value = \
719
self.source._comparison_data(from_entry, path)
720
kind = (from_kind, None)
721
executable = (from_executable, None)
722
changed_content = True
723
# the parent's path is necessarily known at this point.
724
yield(file_id, to_path, changed_content, versioned, parent,
725
name, kind, executable)
728
# This was deprecated before 0.12, but did not have an official warning
729
@symbol_versioning.deprecated_function(symbol_versioning.zero_twelve)
730
def RevisionTree(*args, **kwargs):
731
"""RevisionTree has moved to bzrlib.revisiontree.RevisionTree()
733
Accessing it as bzrlib.tree.RevisionTree has been deprecated as of
736
from bzrlib.revisiontree import RevisionTree as _RevisionTree
737
return _RevisionTree(*args, **kwargs)
added
removed
bzrlib/tree.py
