1
# Copyright (C) 2005 Canonical Ltd
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
"""Tree classes, representing directory at point in time.
21
from collections import deque
22
from cStringIO import StringIO
30
from bzrlib.decorators import needs_read_lock
31
from bzrlib.errors import BzrError, BzrCheckError
32
from bzrlib import errors
33
from bzrlib.inventory import Inventory, InventoryFile
34
from bzrlib.inter import InterObject
35
from bzrlib.osutils import fingerprint_file
36
import bzrlib.revision
37
from bzrlib.trace import mutter, note
41
"""Abstract file tree.
43
There are several subclasses:
45
* `WorkingTree` exists as files on disk editable by the user.
47
* `RevisionTree` is a tree as recorded at some point in the past.
49
Trees contain an `Inventory` object, and also know how to retrieve
50
file texts mentioned in the inventory, either from a working
51
directory or from a store.
53
It is possible for trees to contain files that are not described
54
in their inventory or vice versa; for this use `filenames()`.
56
Trees can be compared, etc, regardless of whether they are working
57
trees or versioned trees.
60
def changes_from(self, other, want_unchanged=False, specific_files=None,
61
extra_trees=None, require_versioned=False, include_root=False,
62
want_unversioned=False):
63
"""Return a TreeDelta of the changes from other to this tree.
65
:param other: A tree to compare with.
66
:param specific_files: An optional list of file paths to restrict the
67
comparison to. When mapping filenames to ids, all matches in all
68
trees (including optional extra_trees) are used, and all children of
69
matched directories are included.
70
:param want_unchanged: An optional boolean requesting the inclusion of
71
unchanged entries in the result.
72
:param extra_trees: An optional list of additional trees to use when
73
mapping the contents of specific_files (paths) to file_ids.
74
:param require_versioned: An optional boolean (defaults to False). When
75
supplied and True all the 'specific_files' must be versioned, or
76
a PathsNotVersionedError will be thrown.
77
:param want_unversioned: Scan for unversioned paths.
79
The comparison will be performed by an InterTree object looked up on
82
# Martin observes that Tree.changes_from returns a TreeDelta and this
83
# may confuse people, because the class name of the returned object is
84
# a synonym of the object referenced in the method name.
85
return InterTree.get(other, self).compare(
86
want_unchanged=want_unchanged,
87
specific_files=specific_files,
88
extra_trees=extra_trees,
89
require_versioned=require_versioned,
90
include_root=include_root,
91
want_unversioned=want_unversioned,
94
def _iter_changes(self, from_tree, include_unchanged=False,
95
specific_files=None, pb=None, extra_trees=None,
96
require_versioned=True, want_unversioned=False):
97
intertree = InterTree.get(from_tree, self)
98
return intertree._iter_changes(include_unchanged, specific_files, pb,
99
extra_trees, require_versioned, want_unversioned=want_unversioned)
102
"""Get a list of the conflicts in the tree.
104
Each conflict is an instance of bzrlib.conflicts.Conflict.
109
"""For trees that can have unversioned files, return all such paths."""
112
def get_parent_ids(self):
113
"""Get the parent ids for this tree.
115
:return: a list of parent ids. [] is returned to indicate
116
a tree with no parents.
117
:raises: BzrError if the parents are not known.
119
raise NotImplementedError(self.get_parent_ids)
121
def has_filename(self, filename):
122
"""True if the tree has given filename."""
123
raise NotImplementedError()
125
def has_id(self, file_id):
126
file_id = osutils.safe_file_id(file_id)
127
return self.inventory.has_id(file_id)
129
__contains__ = has_id
131
def has_or_had_id(self, file_id):
132
file_id = osutils.safe_file_id(file_id)
133
if file_id == self.inventory.root.file_id:
135
return self.inventory.has_id(file_id)
137
def is_ignored(self, filename):
138
"""Check whether the filename is ignored by this tree.
140
:param filename: The relative filename within the tree.
141
:return: True if the filename is ignored.
146
return iter(self.inventory)
148
def id2path(self, file_id):
149
file_id = osutils.safe_file_id(file_id)
150
return self.inventory.id2path(file_id)
152
def is_control_filename(self, filename):
153
"""True if filename is the name of a control file in this tree.
155
:param filename: A filename within the tree. This is a relative path
156
from the root of this tree.
158
This is true IF and ONLY IF the filename is part of the meta data
159
that bzr controls in this tree. I.E. a random .bzr directory placed
160
on disk will not be a control file for this tree.
162
return self.bzrdir.is_control_filename(filename)
165
def iter_entries_by_dir(self, specific_file_ids=None):
166
"""Walk the tree in 'by_dir' order.
168
This will yield each entry in the tree as a (path, entry) tuple. The
169
order that they are yielded is: the contents of a directory are
170
preceeded by the parent of a directory, and all the contents of a
171
directory are grouped together.
173
return self.inventory.iter_entries_by_dir(
174
specific_file_ids=specific_file_ids)
176
def kind(self, file_id):
177
raise NotImplementedError("subclasses must implement kind")
179
def _comparison_data(self, entry, path):
180
"""Return a tuple of kind, executable, stat_value for a file.
182
entry may be None if there is no inventory entry for the file, but
183
path must always be supplied.
185
kind is None if there is no file present (even if an inventory id is
186
present). executable is False for non-file entries.
188
raise NotImplementedError(self._comparison_data)
190
def _file_size(self, entry, stat_value):
191
raise NotImplementedError(self._file_size)
193
def _get_inventory(self):
194
return self._inventory
196
def get_file(self, file_id):
197
"""Return a file object for the file file_id in the tree."""
198
raise NotImplementedError(self.get_file)
200
def get_file_mtime(self, file_id, path=None):
201
"""Return the modification time for a file.
203
:param file_id: The handle for this file.
204
:param path: The path that this file can be found at.
205
These must point to the same object.
207
raise NotImplementedError(self.get_file_mtime)
209
def get_file_by_path(self, path):
210
return self.get_file(self._inventory.path2id(path))
212
def get_symlink_target(self, file_id):
213
"""Get the target for a given file_id.
215
It is assumed that the caller already knows that file_id is referencing
217
:param file_id: Handle for the symlink entry.
218
:return: The path the symlink points to.
220
raise NotImplementedError(self.get_symlink_target)
222
def annotate_iter(self, file_id):
223
"""Return an iterator of revision_id, line tuples
225
For working trees (and mutable trees in general), the special
226
revision_id 'current:' will be used for lines that are new in this
227
tree, e.g. uncommitted changes.
228
:param file_id: The file to produce an annotated version from
230
raise NotImplementedError(self.annotate_iter)
232
inventory = property(_get_inventory,
233
doc="Inventory of this Tree")
235
def _check_retrieved(self, ie, f):
238
fp = fingerprint_file(f)
241
if ie.text_size is not None:
242
if ie.text_size != fp['size']:
243
raise BzrError("mismatched size for file %r in %r" % (ie.file_id, self._store),
244
["inventory expects %d bytes" % ie.text_size,
245
"file is actually %d bytes" % fp['size'],
246
"store is probably damaged/corrupt"])
248
if ie.text_sha1 != fp['sha1']:
249
raise BzrError("wrong SHA-1 for file %r in %r" % (ie.file_id, self._store),
250
["inventory expects %s" % ie.text_sha1,
251
"file is actually %s" % fp['sha1'],
252
"store is probably damaged/corrupt"])
254
def path2id(self, path):
255
"""Return the id for path in this tree."""
256
return self._inventory.path2id(path)
258
def paths2ids(self, paths, trees=[], require_versioned=True):
259
"""Return all the ids that can be reached by walking from paths.
261
Each path is looked up in each this tree and any extras provided in
262
trees, and this is repeated recursively: the children in an extra tree
263
of a directory that has been renamed under a provided path in this tree
264
are all returned, even if none exist until a provided path in this
265
tree, and vice versa.
267
:param paths: An iterable of paths to start converting to ids from.
268
Alternatively, if paths is None, no ids should be calculated and None
269
will be returned. This is offered to make calling the api unconditional
270
for code that *might* take a list of files.
271
:param trees: Additional trees to consider.
272
:param require_versioned: If False, do not raise NotVersionedError if
273
an element of paths is not versioned in this tree and all of trees.
275
return find_ids_across_trees(paths, [self] + list(trees), require_versioned)
277
def print_file(self, file_id):
278
"""Print file with id `file_id` to stdout."""
279
file_id = osutils.safe_file_id(file_id)
281
sys.stdout.write(self.get_file_text(file_id))
286
def revision_tree(self, revision_id):
287
"""Obtain a revision tree for the revision revision_id.
289
The intention of this method is to allow access to possibly cached
290
tree data. Implementors of this method should raise NoSuchRevision if
291
the tree is not locally available, even if they could obtain the
292
tree via a repository or some other means. Callers are responsible
293
for finding the ultimate source for a revision tree.
295
:param revision_id: The revision_id of the requested tree.
297
:raises: NoSuchRevision if the tree cannot be obtained.
299
raise errors.NoSuchRevisionInTree(self, revision_id)
302
"""What files are present in this tree and unknown.
304
:return: an iterator over the unknown files.
311
def filter_unversioned_files(self, paths):
312
"""Filter out paths that are versioned.
314
:return: set of paths.
316
# NB: we specifically *don't* call self.has_filename, because for
317
# WorkingTrees that can indicate files that exist on disk but that
319
pred = self.inventory.has_filename
320
return set((p for p in paths if not pred(p)))
322
def walkdirs(self, prefix=""):
323
"""Walk the contents of this tree from path down.
325
This yields all the data about the contents of a directory at a time.
326
After each directory has been yielded, if the caller has mutated the
327
list to exclude some directories, they are then not descended into.
329
The data yielded is of the form:
330
((directory-relpath, directory-path-from-root, directory-fileid),
331
[(relpath, basename, kind, lstat, path_from_tree_root, file_id,
332
versioned_kind), ...]),
333
- directory-relpath is the containing dirs relpath from prefix
334
- directory-path-from-root is the containing dirs path from /
335
- directory-fileid is the id of the directory if it is versioned.
336
- relpath is the relative path within the subtree being walked.
337
- basename is the basename
338
- kind is the kind of the file now. If unknonwn then the file is not
339
present within the tree - but it may be recorded as versioned. See
341
- lstat is the stat data *if* the file was statted.
342
- path_from_tree_root is the path from the root of the tree.
343
- file_id is the file_id is the entry is versioned.
344
- versioned_kind is the kind of the file as last recorded in the
345
versioning system. If 'unknown' the file is not versioned.
346
One of 'kind' and 'versioned_kind' must not be 'unknown'.
348
:param prefix: Start walking from prefix within the tree rather than
349
at the root. This allows one to walk a subtree but get paths that are
350
relative to a tree rooted higher up.
351
:return: an iterator over the directory data.
353
raise NotImplementedError(self.walkdirs)
356
class EmptyTree(Tree):
359
self._inventory = Inventory(root_id=None)
360
symbol_versioning.warn('EmptyTree is deprecated as of bzr 0.9 please'
361
' use repository.revision_tree instead.',
362
DeprecationWarning, stacklevel=2)
364
def get_parent_ids(self):
367
def get_symlink_target(self, file_id):
370
def has_filename(self, filename):
373
def kind(self, file_id):
374
file_id = osutils.safe_file_id(file_id)
375
assert self._inventory[file_id].kind == "directory"
378
def list_files(self, include_root=False):
381
def __contains__(self, file_id):
382
file_id = osutils.safe_file_id(file_id)
383
return (file_id in self._inventory)
385
def get_file_sha1(self, file_id, path=None, stat_value=None):
389
######################################################################
392
# TODO: Merge these two functions into a single one that can operate
393
# on either a whole tree or a set of files.
395
# TODO: Return the diff in order by filename, not by category or in
396
# random order. Can probably be done by lock-stepping through the
397
# filenames from both trees.
400
def file_status(filename, old_tree, new_tree):
401
"""Return single-letter status, old and new names for a file.
403
The complexity here is in deciding how to represent renames;
404
many complex cases are possible.
406
old_inv = old_tree.inventory
407
new_inv = new_tree.inventory
408
new_id = new_inv.path2id(filename)
409
old_id = old_inv.path2id(filename)
411
if not new_id and not old_id:
412
# easy: doesn't exist in either; not versioned at all
413
if new_tree.is_ignored(filename):
414
return 'I', None, None
416
return '?', None, None
418
# There is now a file of this name, great.
421
# There is no longer a file of this name, but we can describe
422
# what happened to the file that used to have
423
# this name. There are two possibilities: either it was
424
# deleted entirely, or renamed.
426
if new_inv.has_id(old_id):
427
return 'X', old_inv.id2path(old_id), new_inv.id2path(old_id)
429
return 'D', old_inv.id2path(old_id), None
431
# if the file_id is new in this revision, it is added
432
if new_id and not old_inv.has_id(new_id):
435
# if there used to be a file of this name, but that ID has now
436
# disappeared, it is deleted
437
if old_id and not new_inv.has_id(old_id):
444
def find_renames(old_inv, new_inv):
445
for file_id in old_inv:
446
if file_id not in new_inv:
448
old_name = old_inv.id2path(file_id)
449
new_name = new_inv.id2path(file_id)
450
if old_name != new_name:
451
yield (old_name, new_name)
454
def find_ids_across_trees(filenames, trees, require_versioned=True):
455
"""Find the ids corresponding to specified filenames.
457
All matches in all trees will be used, and all children of matched
458
directories will be used.
460
:param filenames: The filenames to find file_ids for (if None, returns
462
:param trees: The trees to find file_ids within
463
:param require_versioned: if true, all specified filenames must occur in
465
:return: a set of file ids for the specified filenames and their children.
469
specified_path_ids = _find_ids_across_trees(filenames, trees,
471
return _find_children_across_trees(specified_path_ids, trees)
474
def _find_ids_across_trees(filenames, trees, require_versioned):
475
"""Find the ids corresponding to specified filenames.
477
All matches in all trees will be used, but subdirectories are not scanned.
479
:param filenames: The filenames to find file_ids for
480
:param trees: The trees to find file_ids within
481
:param require_versioned: if true, all specified filenames must occur in
483
:return: a set of (path, file ids) for the specified filenames
486
interesting_ids = set()
487
for tree_path in filenames:
490
file_id = tree.path2id(tree_path)
491
if file_id is not None:
492
interesting_ids.add(file_id)
495
not_versioned.append(tree_path)
496
if len(not_versioned) > 0 and require_versioned:
497
raise errors.PathsNotVersionedError(not_versioned)
498
return interesting_ids
501
def _find_children_across_trees(specified_ids, trees):
502
"""Return a set including specified ids and their children
504
All matches in all trees will be used.
506
:param trees: The trees to find file_ids within
507
:return: a set containing all specified ids and their children
509
interesting_ids = set(specified_ids)
510
pending = interesting_ids
511
# now handle children of interesting ids
512
# we loop so that we handle all children of each id in both trees
513
while len(pending) > 0:
515
for file_id in pending:
517
if not tree.has_id(file_id):
519
entry = tree.inventory[file_id]
520
for child in getattr(entry, 'children', {}).itervalues():
521
if child.file_id not in interesting_ids:
522
new_pending.add(child.file_id)
523
interesting_ids.update(new_pending)
524
pending = new_pending
525
return interesting_ids
528
class InterTree(InterObject):
529
"""This class represents operations taking place between two Trees.
531
Its instances have methods like 'compare' and contain references to the
532
source and target trees these operations are to be carried out on.
534
clients of bzrlib should not need to use InterTree directly, rather they
535
should use the convenience methods on Tree such as 'Tree.compare()' which
536
will pass through to InterTree as appropriate.
542
def compare(self, want_unchanged=False, specific_files=None,
543
extra_trees=None, require_versioned=False, include_root=False,
544
want_unversioned=False):
545
"""Return the changes from source to target.
547
:return: A TreeDelta.
548
:param specific_files: An optional list of file paths to restrict the
549
comparison to. When mapping filenames to ids, all matches in all
550
trees (including optional extra_trees) are used, and all children of
551
matched directories are included.
552
:param want_unchanged: An optional boolean requesting the inclusion of
553
unchanged entries in the result.
554
:param extra_trees: An optional list of additional trees to use when
555
mapping the contents of specific_files (paths) to file_ids.
556
:param require_versioned: An optional boolean (defaults to False). When
557
supplied and True all the 'specific_files' must be versioned, or
558
a PathsNotVersionedError will be thrown.
559
:param want_unversioned: Scan for unversioned paths.
561
# NB: show_status depends on being able to pass in non-versioned files
562
# and report them as unknown
563
trees = (self.source,)
564
if extra_trees is not None:
565
trees = trees + tuple(extra_trees)
566
# target is usually the newer tree:
567
specific_file_ids = self.target.paths2ids(specific_files, trees,
568
require_versioned=require_versioned)
569
if specific_files and not specific_file_ids:
570
# All files are unversioned, so just return an empty delta
571
# _compare_trees would think we want a complete delta
572
result = delta.TreeDelta()
573
fake_entry = InventoryFile('unused', 'unused', 'unused')
574
result.unversioned = [(path, None,
575
self.target._comparison_data(fake_entry, path)[0]) for path in
578
return delta._compare_trees(self.source, self.target, want_unchanged,
579
specific_files, include_root, extra_trees=extra_trees,
580
want_unversioned=want_unversioned)
582
def _iter_changes(self, include_unchanged=False,
583
specific_files=None, pb=None, extra_trees=[],
584
require_versioned=True, want_unversioned=False):
585
"""Generate an iterator of changes between trees.
588
(file_id, (path_in_source, path_in_target),
589
changed_content, versioned, parent, name, kind,
592
Changed_content is True if the file's content has changed. This
593
includes changes to its kind, and to a symlink's target.
595
versioned, parent, name, kind, executable are tuples of (from, to).
596
If a file is missing in a tree, its kind is None.
598
Iteration is done in parent-to-child order, relative to the target
601
There is no guarantee that all paths are in sorted order: the
602
requirement to expand the search due to renames may result in children
603
that should be found early being found late in the search, after
604
lexically later results have been returned.
605
:param require_versioned: Raise errors.PathsNotVersionedError if a
606
path in the specific_files list is not versioned in one of
607
source, target or extra_trees.
608
:param want_unversioned: Should unversioned files be returned in the
609
output. An unversioned file is defined as one with (False, False)
610
for the versioned pair.
612
lookup_trees = [self.source]
614
lookup_trees.extend(extra_trees)
615
specific_file_ids = self.target.paths2ids(specific_files,
616
lookup_trees, require_versioned=require_versioned)
618
all_unversioned = sorted([(p.split('/'), p) for p in self.target.extras()
619
if not specific_files or
620
osutils.is_inside_any(specific_files, p)])
621
all_unversioned = deque(all_unversioned)
623
all_unversioned = deque()
625
from_entries_by_dir = list(self.source.inventory.iter_entries_by_dir(
626
specific_file_ids=specific_file_ids))
627
from_data = dict((e.file_id, (p, e)) for p, e in from_entries_by_dir)
628
to_entries_by_dir = list(self.target.inventory.iter_entries_by_dir(
629
specific_file_ids=specific_file_ids))
630
num_entries = len(from_entries_by_dir) + len(to_entries_by_dir)
632
# the unversioned path lookup only occurs on real trees - where there
633
# can be extras. So the fake_entry is solely used to look up
634
# executable it values when execute is not supported.
635
fake_entry = InventoryFile('unused', 'unused', 'unused')
636
for to_path, to_entry in to_entries_by_dir:
637
while all_unversioned and all_unversioned[0][0] < to_path.split('/'):
638
unversioned_path = all_unversioned.popleft()
639
to_kind, to_executable, to_stat = \
640
self.target._comparison_data(fake_entry, unversioned_path[1])
641
yield (None, (None, unversioned_path[1]), True, (False, False),
643
(None, unversioned_path[0][-1]),
645
(None, to_executable))
646
file_id = to_entry.file_id
647
to_paths[file_id] = to_path
649
changed_content = False
650
from_path, from_entry = from_data.get(file_id, (None, None))
651
from_versioned = (from_entry is not None)
652
if from_entry is not None:
653
from_versioned = True
654
from_name = from_entry.name
655
from_parent = from_entry.parent_id
656
from_kind, from_executable, from_stat = \
657
self.source._comparison_data(from_entry, from_path)
660
from_versioned = False
664
from_executable = None
665
versioned = (from_versioned, True)
666
to_kind, to_executable, to_stat = \
667
self.target._comparison_data(to_entry, to_path)
668
kind = (from_kind, to_kind)
669
if kind[0] != kind[1]:
670
changed_content = True
671
elif from_kind == 'file':
672
from_size = self.source._file_size(from_entry, from_stat)
673
to_size = self.target._file_size(to_entry, to_stat)
674
if from_size != to_size:
675
changed_content = True
676
elif (self.source.get_file_sha1(file_id, from_path, from_stat) !=
677
self.target.get_file_sha1(file_id, to_path, to_stat)):
678
changed_content = True
679
elif from_kind == 'symlink':
680
if (self.source.get_symlink_target(file_id) !=
681
self.target.get_symlink_target(file_id)):
682
changed_content = True
683
parent = (from_parent, to_entry.parent_id)
684
name = (from_name, to_entry.name)
685
executable = (from_executable, to_executable)
687
pb.update('comparing files', entry_count, num_entries)
688
if (changed_content is not False or versioned[0] != versioned[1]
689
or parent[0] != parent[1] or name[0] != name[1] or
690
executable[0] != executable[1] or include_unchanged):
691
yield (file_id, (from_path, to_path), changed_content,
692
versioned, parent, name, kind, executable)
694
while all_unversioned:
695
# yield any trailing unversioned paths
696
unversioned_path = all_unversioned.popleft()
697
to_kind, to_executable, to_stat = \
698
self.target._comparison_data(fake_entry, unversioned_path[1])
699
yield (None, (None, unversioned_path[1]), True, (False, False),
701
(None, unversioned_path[0][-1]),
703
(None, to_executable))
705
def get_to_path(to_entry):
706
if to_entry.parent_id is None:
707
to_path = '' # the root
709
if to_entry.parent_id not in to_paths:
711
return get_to_path(self.target.inventory[to_entry.parent_id])
712
to_path = osutils.pathjoin(to_paths[to_entry.parent_id],
714
to_paths[to_entry.file_id] = to_path
717
for path, from_entry in from_entries_by_dir:
718
file_id = from_entry.file_id
719
if file_id in to_paths:
722
if not file_id in self.target.inventory:
723
# common case - paths we have not emitted are not present in
727
to_path = get_to_path(self.target.inventory[file_id])
730
pb.update('comparing files', entry_count, num_entries)
731
versioned = (True, False)
732
parent = (from_entry.parent_id, None)
733
name = (from_entry.name, None)
734
from_kind, from_executable, stat_value = \
735
self.source._comparison_data(from_entry, path)
736
kind = (from_kind, None)
737
executable = (from_executable, None)
738
changed_content = True
739
# the parent's path is necessarily known at this point.
740
yield(file_id, (path, to_path), changed_content, versioned, parent,
741
name, kind, executable)
744
# This was deprecated before 0.12, but did not have an official warning
745
@symbol_versioning.deprecated_function(symbol_versioning.zero_twelve)
746
def RevisionTree(*args, **kwargs):
747
"""RevisionTree has moved to bzrlib.revisiontree.RevisionTree()
749
Accessing it as bzrlib.tree.RevisionTree has been deprecated as of
752
from bzrlib.revisiontree import RevisionTree as _RevisionTree
753
return _RevisionTree(*args, **kwargs)
added
removed
bzrlib/tree.py
