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
26
conflicts as _mod_conflicts,
29
revision as _mod_revision,
33
from bzrlib.decorators import needs_read_lock
34
from bzrlib.errors import BzrError, BzrCheckError
35
from bzrlib import errors
36
from bzrlib.inventory import Inventory, InventoryFile
37
from bzrlib.inter import InterObject
38
from bzrlib.osutils import fingerprint_file
39
import bzrlib.revision
40
from bzrlib.symbol_versioning import deprecated_function, deprecated_in
41
from bzrlib.trace import mutter, note
45
"""Abstract file tree.
47
There are several subclasses:
49
* `WorkingTree` exists as files on disk editable by the user.
51
* `RevisionTree` is a tree as recorded at some point in the past.
53
Trees contain an `Inventory` object, and also know how to retrieve
54
file texts mentioned in the inventory, either from a working
55
directory or from a store.
57
It is possible for trees to contain files that are not described
58
in their inventory or vice versa; for this use `filenames()`.
60
Trees can be compared, etc, regardless of whether they are working
61
trees or versioned trees.
64
def changes_from(self, other, want_unchanged=False, specific_files=None,
65
extra_trees=None, require_versioned=False, include_root=False,
66
want_unversioned=False):
67
"""Return a TreeDelta of the changes from other to this tree.
69
:param other: A tree to compare with.
70
:param specific_files: An optional list of file paths to restrict the
71
comparison to. When mapping filenames to ids, all matches in all
72
trees (including optional extra_trees) are used, and all children of
73
matched directories are included.
74
:param want_unchanged: An optional boolean requesting the inclusion of
75
unchanged entries in the result.
76
:param extra_trees: An optional list of additional trees to use when
77
mapping the contents of specific_files (paths) to file_ids.
78
:param require_versioned: An optional boolean (defaults to False). When
79
supplied and True all the 'specific_files' must be versioned, or
80
a PathsNotVersionedError will be thrown.
81
:param want_unversioned: Scan for unversioned paths.
83
The comparison will be performed by an InterTree object looked up on
86
# Martin observes that Tree.changes_from returns a TreeDelta and this
87
# may confuse people, because the class name of the returned object is
88
# a synonym of the object referenced in the method name.
89
return InterTree.get(other, self).compare(
90
want_unchanged=want_unchanged,
91
specific_files=specific_files,
92
extra_trees=extra_trees,
93
require_versioned=require_versioned,
94
include_root=include_root,
95
want_unversioned=want_unversioned,
98
@symbol_versioning.deprecated_method(symbol_versioning.one_three)
99
def _iter_changes(self, *args, **kwargs):
100
return self.iter_changes(*args, **kwargs)
102
def iter_changes(self, from_tree, include_unchanged=False,
103
specific_files=None, pb=None, extra_trees=None,
104
require_versioned=True, want_unversioned=False):
105
intertree = InterTree.get(from_tree, self)
106
return intertree.iter_changes(include_unchanged, specific_files, pb,
107
extra_trees, require_versioned, want_unversioned=want_unversioned)
110
"""Get a list of the conflicts in the tree.
112
Each conflict is an instance of bzrlib.conflicts.Conflict.
114
return _mod_conflicts.ConflictList()
117
"""For trees that can have unversioned files, return all such paths."""
120
def get_parent_ids(self):
121
"""Get the parent ids for this tree.
123
:return: a list of parent ids. [] is returned to indicate
124
a tree with no parents.
125
:raises: BzrError if the parents are not known.
127
raise NotImplementedError(self.get_parent_ids)
129
def has_filename(self, filename):
130
"""True if the tree has given filename."""
131
raise NotImplementedError(self.has_filename)
133
def has_id(self, file_id):
134
return self.inventory.has_id(file_id)
136
def __contains__(self, file_id):
137
return self.has_id(file_id)
139
def has_or_had_id(self, file_id):
140
if file_id == self.inventory.root.file_id:
142
return self.inventory.has_id(file_id)
144
def is_ignored(self, filename):
145
"""Check whether the filename is ignored by this tree.
147
:param filename: The relative filename within the tree.
148
:return: True if the filename is ignored.
153
return iter(self.inventory)
155
def all_file_ids(self):
156
"""Iterate through all file ids, including ids for missing files."""
157
return set(self.inventory)
159
def id2path(self, file_id):
160
"""Return the path for a file id.
164
return self.inventory.id2path(file_id)
166
def is_control_filename(self, filename):
167
"""True if filename is the name of a control file in this tree.
169
:param filename: A filename within the tree. This is a relative path
170
from the root of this tree.
172
This is true IF and ONLY IF the filename is part of the meta data
173
that bzr controls in this tree. I.E. a random .bzr directory placed
174
on disk will not be a control file for this tree.
176
return self.bzrdir.is_control_filename(filename)
179
def iter_entries_by_dir(self, specific_file_ids=None):
180
"""Walk the tree in 'by_dir' order.
182
This will yield each entry in the tree as a (path, entry) tuple.
183
The order that they are yielded is:
185
Directories are walked in a depth-first lexicographical order,
186
however, whenever a directory is reached, all of its direct child
187
nodes are yielded in lexicographical order before yielding the
190
For example, in the tree::
200
The yield order (ignoring root) would be::
201
a, f, a/b, a/d, a/b/c, a/d/e, f/g
203
return self.inventory.iter_entries_by_dir(
204
specific_file_ids=specific_file_ids)
206
def iter_references(self):
207
for path, entry in self.iter_entries_by_dir():
208
if entry.kind == 'tree-reference':
209
yield path, entry.file_id
211
def kind(self, file_id):
212
raise NotImplementedError("Tree subclass %s must implement kind"
213
% self.__class__.__name__)
215
def stored_kind(self, file_id):
216
"""File kind stored for this file_id.
218
May not match kind on disk for working trees. Always available
219
for versioned files, even when the file itself is missing.
221
return self.kind(file_id)
223
def path_content_summary(self, path):
224
"""Get a summary of the information about path.
226
:param path: A relative path within the tree.
227
:return: A tuple containing kind, size, exec, sha1-or-link.
228
Kind is always present (see tree.kind()).
229
size is present if kind is file, None otherwise.
230
exec is None unless kind is file and the platform supports the 'x'
232
sha1-or-link is the link target if kind is symlink, or the sha1 if
233
it can be obtained without reading the file.
235
raise NotImplementedError(self.path_content_summary)
237
def get_reference_revision(self, file_id, path=None):
238
raise NotImplementedError("Tree subclass %s must implement "
239
"get_reference_revision"
240
% self.__class__.__name__)
242
def _comparison_data(self, entry, path):
243
"""Return a tuple of kind, executable, stat_value for a file.
245
entry may be None if there is no inventory entry for the file, but
246
path must always be supplied.
248
kind is None if there is no file present (even if an inventory id is
249
present). executable is False for non-file entries.
251
raise NotImplementedError(self._comparison_data)
253
def _file_size(self, entry, stat_value):
254
raise NotImplementedError(self._file_size)
256
def _get_inventory(self):
257
return self._inventory
259
def get_file(self, file_id, path=None):
260
"""Return a file object for the file file_id in the tree.
262
If both file_id and path are defined, it is implementation defined as
263
to which one is used.
265
raise NotImplementedError(self.get_file)
267
def get_file_text(self, file_id, path=None):
268
"""Return the byte content of a file.
270
:param file_id: The file_id of the file.
271
:param path: The path of the file.
272
If both file_id and path are supplied, an implementation may use
275
my_file = self.get_file(file_id, path)
277
return my_file.read()
281
def get_file_lines(self, file_id, path=None):
282
"""Return the content of a file, as lines.
284
:param file_id: The file_id of the file.
285
:param path: The path of the file.
286
If both file_id and path are supplied, an implementation may use
289
return osutils.split_lines(self.get_file_text(file_id, path))
291
def get_file_mtime(self, file_id, path=None):
292
"""Return the modification time for a file.
294
:param file_id: The handle for this file.
295
:param path: The path that this file can be found at.
296
These must point to the same object.
298
raise NotImplementedError(self.get_file_mtime)
300
def get_file_size(self, file_id):
301
"""Return the size of a file in bytes.
303
This applies only to regular files. If invoked on directories or
304
symlinks, it will return None.
305
:param file_id: The file-id of the file
307
raise NotImplementedError(self.get_file_size)
309
def get_file_by_path(self, path):
310
return self.get_file(self._inventory.path2id(path), path)
312
def iter_files_bytes(self, desired_files):
313
"""Iterate through file contents.
315
Files will not necessarily be returned in the order they occur in
316
desired_files. No specific order is guaranteed.
318
Yields pairs of identifier, bytes_iterator. identifier is an opaque
319
value supplied by the caller as part of desired_files. It should
320
uniquely identify the file version in the caller's context. (Examples:
321
an index number or a TreeTransform trans_id.)
323
bytes_iterator is an iterable of bytestrings for the file. The
324
kind of iterable and length of the bytestrings are unspecified, but for
325
this implementation, it is a tuple containing a single bytestring with
326
the complete text of the file.
328
:param desired_files: a list of (file_id, identifier) pairs
330
for file_id, identifier in desired_files:
331
# We wrap the string in a tuple so that we can return an iterable
332
# of bytestrings. (Technically, a bytestring is also an iterable
333
# of bytestrings, but iterating through each character is not
335
cur_file = (self.get_file_text(file_id),)
336
yield identifier, cur_file
338
def get_symlink_target(self, file_id):
339
"""Get the target for a given file_id.
341
It is assumed that the caller already knows that file_id is referencing
343
:param file_id: Handle for the symlink entry.
344
:return: The path the symlink points to.
346
raise NotImplementedError(self.get_symlink_target)
348
def get_root_id(self):
349
"""Return the file_id for the root of this tree."""
350
raise NotImplementedError(self.get_root_id)
352
def annotate_iter(self, file_id,
353
default_revision=_mod_revision.CURRENT_REVISION):
354
"""Return an iterator of revision_id, line tuples.
356
For working trees (and mutable trees in general), the special
357
revision_id 'current:' will be used for lines that are new in this
358
tree, e.g. uncommitted changes.
359
:param file_id: The file to produce an annotated version from
360
:param default_revision: For lines that don't match a basis, mark them
361
with this revision id. Not all implementations will make use of
364
raise NotImplementedError(self.annotate_iter)
366
def _get_plan_merge_data(self, file_id, other, base):
367
from bzrlib import merge, versionedfile
368
vf = versionedfile._PlanMergeVersionedFile(file_id)
369
last_revision_a = self._get_file_revision(file_id, vf, 'this:')
370
last_revision_b = other._get_file_revision(file_id, vf, 'other:')
372
last_revision_base = None
374
last_revision_base = base._get_file_revision(file_id, vf, 'base:')
375
return vf, last_revision_a, last_revision_b, last_revision_base
377
def plan_file_merge(self, file_id, other, base=None):
378
"""Generate a merge plan based on annotations.
380
If the file contains uncommitted changes in this tree, they will be
381
attributed to the 'current:' pseudo-revision. If the file contains
382
uncommitted changes in the other tree, they will be assigned to the
383
'other:' pseudo-revision.
385
data = self._get_plan_merge_data(file_id, other, base)
386
vf, last_revision_a, last_revision_b, last_revision_base = data
387
return vf.plan_merge(last_revision_a, last_revision_b,
390
def plan_file_lca_merge(self, file_id, other, base=None):
391
"""Generate a merge plan based lca-newness.
393
If the file contains uncommitted changes in this tree, they will be
394
attributed to the 'current:' pseudo-revision. If the file contains
395
uncommitted changes in the other tree, they will be assigned to the
396
'other:' pseudo-revision.
398
data = self._get_plan_merge_data(file_id, other, base)
399
vf, last_revision_a, last_revision_b, last_revision_base = data
400
return vf.plan_lca_merge(last_revision_a, last_revision_b,
403
def _iter_parent_trees(self):
404
"""Iterate through parent trees, defaulting to Tree.revision_tree."""
405
for revision_id in self.get_parent_ids():
407
yield self.revision_tree(revision_id)
408
except errors.NoSuchRevisionInTree:
409
yield self.repository.revision_tree(revision_id)
412
def _file_revision(revision_tree, file_id):
413
"""Determine the revision associated with a file in a given tree."""
414
revision_tree.lock_read()
416
return revision_tree.inventory[file_id].revision
418
revision_tree.unlock()
420
def _get_file_revision(self, file_id, vf, tree_revision):
421
"""Ensure that file_id, tree_revision is in vf to plan the merge."""
423
if getattr(self, '_repository', None) is None:
424
last_revision = tree_revision
425
parent_keys = [(file_id, self._file_revision(t, file_id)) for t in
426
self._iter_parent_trees()]
427
vf.add_lines((file_id, last_revision), parent_keys,
428
self.get_file(file_id).readlines())
429
repo = self.branch.repository
432
last_revision = self._file_revision(self, file_id)
433
base_vf = self._repository.texts
434
if base_vf not in vf.fallback_versionedfiles:
435
vf.fallback_versionedfiles.append(base_vf)
438
inventory = property(_get_inventory,
439
doc="Inventory of this Tree")
441
def _check_retrieved(self, ie, f):
444
fp = fingerprint_file(f)
447
if ie.text_size is not None:
448
if ie.text_size != fp['size']:
449
raise BzrError("mismatched size for file %r in %r" % (ie.file_id, self._store),
450
["inventory expects %d bytes" % ie.text_size,
451
"file is actually %d bytes" % fp['size'],
452
"store is probably damaged/corrupt"])
454
if ie.text_sha1 != fp['sha1']:
455
raise BzrError("wrong SHA-1 for file %r in %r" % (ie.file_id, self._store),
456
["inventory expects %s" % ie.text_sha1,
457
"file is actually %s" % fp['sha1'],
458
"store is probably damaged/corrupt"])
461
def path2id(self, path):
462
"""Return the id for path in this tree."""
463
return self._inventory.path2id(path)
465
def paths2ids(self, paths, trees=[], require_versioned=True):
466
"""Return all the ids that can be reached by walking from paths.
468
Each path is looked up in this tree and any extras provided in
469
trees, and this is repeated recursively: the children in an extra tree
470
of a directory that has been renamed under a provided path in this tree
471
are all returned, even if none exist under a provided path in this
472
tree, and vice versa.
474
:param paths: An iterable of paths to start converting to ids from.
475
Alternatively, if paths is None, no ids should be calculated and None
476
will be returned. This is offered to make calling the api unconditional
477
for code that *might* take a list of files.
478
:param trees: Additional trees to consider.
479
:param require_versioned: If False, do not raise NotVersionedError if
480
an element of paths is not versioned in this tree and all of trees.
482
return find_ids_across_trees(paths, [self] + list(trees), require_versioned)
484
def iter_children(self, file_id):
485
entry = self.iter_entries_by_dir([file_id]).next()[1]
486
for child in getattr(entry, 'children', {}).itervalues():
489
@symbol_versioning.deprecated_method(symbol_versioning.one_six)
490
def print_file(self, file_id):
491
"""Print file with id `file_id` to stdout."""
493
sys.stdout.write(self.get_file_text(file_id))
498
def revision_tree(self, revision_id):
499
"""Obtain a revision tree for the revision revision_id.
501
The intention of this method is to allow access to possibly cached
502
tree data. Implementors of this method should raise NoSuchRevision if
503
the tree is not locally available, even if they could obtain the
504
tree via a repository or some other means. Callers are responsible
505
for finding the ultimate source for a revision tree.
507
:param revision_id: The revision_id of the requested tree.
509
:raises: NoSuchRevision if the tree cannot be obtained.
511
raise errors.NoSuchRevisionInTree(self, revision_id)
514
"""What files are present in this tree and unknown.
516
:return: an iterator over the unknown files.
523
def filter_unversioned_files(self, paths):
524
"""Filter out paths that are versioned.
526
:return: set of paths.
528
# NB: we specifically *don't* call self.has_filename, because for
529
# WorkingTrees that can indicate files that exist on disk but that
531
pred = self.inventory.has_filename
532
return set((p for p in paths if not pred(p)))
534
def walkdirs(self, prefix=""):
535
"""Walk the contents of this tree from path down.
537
This yields all the data about the contents of a directory at a time.
538
After each directory has been yielded, if the caller has mutated the
539
list to exclude some directories, they are then not descended into.
541
The data yielded is of the form:
542
((directory-relpath, directory-path-from-root, directory-fileid),
543
[(relpath, basename, kind, lstat, path_from_tree_root, file_id,
544
versioned_kind), ...]),
545
- directory-relpath is the containing dirs relpath from prefix
546
- directory-path-from-root is the containing dirs path from /
547
- directory-fileid is the id of the directory if it is versioned.
548
- relpath is the relative path within the subtree being walked.
549
- basename is the basename
550
- kind is the kind of the file now. If unknonwn then the file is not
551
present within the tree - but it may be recorded as versioned. See
553
- lstat is the stat data *if* the file was statted.
554
- path_from_tree_root is the path from the root of the tree.
555
- file_id is the file_id if the entry is versioned.
556
- versioned_kind is the kind of the file as last recorded in the
557
versioning system. If 'unknown' the file is not versioned.
558
One of 'kind' and 'versioned_kind' must not be 'unknown'.
560
:param prefix: Start walking from prefix within the tree rather than
561
at the root. This allows one to walk a subtree but get paths that are
562
relative to a tree rooted higher up.
563
:return: an iterator over the directory data.
565
raise NotImplementedError(self.walkdirs)
567
def supports_content_filtering(self):
570
def iter_search_rules(self, path_names, pref_names=None,
571
_default_searcher=rules._per_user_searcher):
572
"""Find the preferences for filenames in a tree.
574
:param path_names: an iterable of paths to find attributes for.
575
Paths are given relative to the root of the tree.
576
:param pref_names: the list of preferences to lookup - None for all
577
:param _default_searcher: private parameter to assist testing - don't use
578
:return: an iterator of tuple sequences, one per path-name.
579
See _RulesSearcher.get_items for details on the tuple sequence.
581
searcher = self._get_rules_searcher(_default_searcher)
582
if searcher is not None:
583
if pref_names is not None:
584
for path in path_names:
585
yield searcher.get_selected_items(path, pref_names)
587
for path in path_names:
588
yield searcher.get_items(path)
591
def _get_rules_searcher(self, default_searcher):
592
"""Get the RulesSearcher for this tree given the default one."""
593
searcher = default_searcher
597
class EmptyTree(Tree):
600
self._inventory = Inventory(root_id=None)
601
symbol_versioning.warn('EmptyTree is deprecated as of bzr 0.9 please'
602
' use repository.revision_tree instead.',
603
DeprecationWarning, stacklevel=2)
605
def get_parent_ids(self):
608
def get_symlink_target(self, file_id):
611
def has_filename(self, filename):
614
def kind(self, file_id):
617
def list_files(self, include_root=False):
620
def __contains__(self, file_id):
621
return (file_id in self._inventory)
623
def get_file_sha1(self, file_id, path=None, stat_value=None):
627
######################################################################
630
# TODO: Merge these two functions into a single one that can operate
631
# on either a whole tree or a set of files.
633
# TODO: Return the diff in order by filename, not by category or in
634
# random order. Can probably be done by lock-stepping through the
635
# filenames from both trees.
638
def file_status(filename, old_tree, new_tree):
639
"""Return single-letter status, old and new names for a file.
641
The complexity here is in deciding how to represent renames;
642
many complex cases are possible.
644
old_inv = old_tree.inventory
645
new_inv = new_tree.inventory
646
new_id = new_inv.path2id(filename)
647
old_id = old_inv.path2id(filename)
649
if not new_id and not old_id:
650
# easy: doesn't exist in either; not versioned at all
651
if new_tree.is_ignored(filename):
652
return 'I', None, None
654
return '?', None, None
656
# There is now a file of this name, great.
659
# There is no longer a file of this name, but we can describe
660
# what happened to the file that used to have
661
# this name. There are two possibilities: either it was
662
# deleted entirely, or renamed.
663
if new_inv.has_id(old_id):
664
return 'X', old_inv.id2path(old_id), new_inv.id2path(old_id)
666
return 'D', old_inv.id2path(old_id), None
668
# if the file_id is new in this revision, it is added
669
if new_id and not old_inv.has_id(new_id):
672
# if there used to be a file of this name, but that ID has now
673
# disappeared, it is deleted
674
if old_id and not new_inv.has_id(old_id):
680
@deprecated_function(deprecated_in((1, 9, 0)))
681
def find_renames(old_inv, new_inv):
682
for file_id in old_inv:
683
if file_id not in new_inv:
685
old_name = old_inv.id2path(file_id)
686
new_name = new_inv.id2path(file_id)
687
if old_name != new_name:
688
yield (old_name, new_name)
691
def find_ids_across_trees(filenames, trees, require_versioned=True):
692
"""Find the ids corresponding to specified filenames.
694
All matches in all trees will be used, and all children of matched
695
directories will be used.
697
:param filenames: The filenames to find file_ids for (if None, returns
699
:param trees: The trees to find file_ids within
700
:param require_versioned: if true, all specified filenames must occur in
702
:return: a set of file ids for the specified filenames and their children.
706
specified_path_ids = _find_ids_across_trees(filenames, trees,
708
return _find_children_across_trees(specified_path_ids, trees)
711
def _find_ids_across_trees(filenames, trees, require_versioned):
712
"""Find the ids corresponding to specified filenames.
714
All matches in all trees will be used, but subdirectories are not scanned.
716
:param filenames: The filenames to find file_ids for
717
:param trees: The trees to find file_ids within
718
:param require_versioned: if true, all specified filenames must occur in
720
:return: a set of file ids for the specified filenames
723
interesting_ids = set()
724
for tree_path in filenames:
727
file_id = tree.path2id(tree_path)
728
if file_id is not None:
729
interesting_ids.add(file_id)
732
not_versioned.append(tree_path)
733
if len(not_versioned) > 0 and require_versioned:
734
raise errors.PathsNotVersionedError(not_versioned)
735
return interesting_ids
738
def _find_children_across_trees(specified_ids, trees):
739
"""Return a set including specified ids and their children.
741
All matches in all trees will be used.
743
:param trees: The trees to find file_ids within
744
:return: a set containing all specified ids and their children
746
interesting_ids = set(specified_ids)
747
pending = interesting_ids
748
# now handle children of interesting ids
749
# we loop so that we handle all children of each id in both trees
750
while len(pending) > 0:
752
for file_id in pending:
754
if not tree.has_id(file_id):
756
for child_id in tree.iter_children(file_id):
757
if child_id not in interesting_ids:
758
new_pending.add(child_id)
759
interesting_ids.update(new_pending)
760
pending = new_pending
761
return interesting_ids
764
class InterTree(InterObject):
765
"""This class represents operations taking place between two Trees.
767
Its instances have methods like 'compare' and contain references to the
768
source and target trees these operations are to be carried out on.
770
Clients of bzrlib should not need to use InterTree directly, rather they
771
should use the convenience methods on Tree such as 'Tree.compare()' which
772
will pass through to InterTree as appropriate.
778
def compare(self, want_unchanged=False, specific_files=None,
779
extra_trees=None, require_versioned=False, include_root=False,
780
want_unversioned=False):
781
"""Return the changes from source to target.
783
:return: A TreeDelta.
784
:param specific_files: An optional list of file paths to restrict the
785
comparison to. When mapping filenames to ids, all matches in all
786
trees (including optional extra_trees) are used, and all children of
787
matched directories are included.
788
:param want_unchanged: An optional boolean requesting the inclusion of
789
unchanged entries in the result.
790
:param extra_trees: An optional list of additional trees to use when
791
mapping the contents of specific_files (paths) to file_ids.
792
:param require_versioned: An optional boolean (defaults to False). When
793
supplied and True all the 'specific_files' must be versioned, or
794
a PathsNotVersionedError will be thrown.
795
:param want_unversioned: Scan for unversioned paths.
797
# NB: show_status depends on being able to pass in non-versioned files
798
# and report them as unknown
799
trees = (self.source,)
800
if extra_trees is not None:
801
trees = trees + tuple(extra_trees)
802
# target is usually the newer tree:
803
specific_file_ids = self.target.paths2ids(specific_files, trees,
804
require_versioned=require_versioned)
805
if specific_files and not specific_file_ids:
806
# All files are unversioned, so just return an empty delta
807
# _compare_trees would think we want a complete delta
808
result = delta.TreeDelta()
809
fake_entry = InventoryFile('unused', 'unused', 'unused')
810
result.unversioned = [(path, None,
811
self.target._comparison_data(fake_entry, path)[0]) for path in
814
return delta._compare_trees(self.source, self.target, want_unchanged,
815
specific_files, include_root, extra_trees=extra_trees,
816
require_versioned=require_versioned,
817
want_unversioned=want_unversioned)
819
def iter_changes(self, include_unchanged=False,
820
specific_files=None, pb=None, extra_trees=[],
821
require_versioned=True, want_unversioned=False):
822
"""Generate an iterator of changes between trees.
825
(file_id, (path_in_source, path_in_target),
826
changed_content, versioned, parent, name, kind,
829
Changed_content is True if the file's content has changed. This
830
includes changes to its kind, and to a symlink's target.
832
versioned, parent, name, kind, executable are tuples of (from, to).
833
If a file is missing in a tree, its kind is None.
835
Iteration is done in parent-to-child order, relative to the target
838
There is no guarantee that all paths are in sorted order: the
839
requirement to expand the search due to renames may result in children
840
that should be found early being found late in the search, after
841
lexically later results have been returned.
842
:param require_versioned: Raise errors.PathsNotVersionedError if a
843
path in the specific_files list is not versioned in one of
844
source, target or extra_trees.
845
:param want_unversioned: Should unversioned files be returned in the
846
output. An unversioned file is defined as one with (False, False)
847
for the versioned pair.
850
lookup_trees = [self.source]
852
lookup_trees.extend(extra_trees)
853
if specific_files == []:
854
specific_file_ids = []
856
specific_file_ids = self.target.paths2ids(specific_files,
857
lookup_trees, require_versioned=require_versioned)
859
all_unversioned = sorted([(p.split('/'), p) for p in
861
if specific_files is None or
862
osutils.is_inside_any(specific_files, p)])
863
all_unversioned = deque(all_unversioned)
865
all_unversioned = deque()
867
from_entries_by_dir = list(self.source.iter_entries_by_dir(
868
specific_file_ids=specific_file_ids))
869
from_data = dict((e.file_id, (p, e)) for p, e in from_entries_by_dir)
870
to_entries_by_dir = list(self.target.iter_entries_by_dir(
871
specific_file_ids=specific_file_ids))
872
num_entries = len(from_entries_by_dir) + len(to_entries_by_dir)
874
# the unversioned path lookup only occurs on real trees - where there
875
# can be extras. So the fake_entry is solely used to look up
876
# executable it values when execute is not supported.
877
fake_entry = InventoryFile('unused', 'unused', 'unused')
878
for to_path, to_entry in to_entries_by_dir:
879
while all_unversioned and all_unversioned[0][0] < to_path.split('/'):
880
unversioned_path = all_unversioned.popleft()
881
to_kind, to_executable, to_stat = \
882
self.target._comparison_data(fake_entry, unversioned_path[1])
883
yield (None, (None, unversioned_path[1]), True, (False, False),
885
(None, unversioned_path[0][-1]),
887
(None, to_executable))
888
file_id = to_entry.file_id
889
to_paths[file_id] = to_path
891
changed_content = False
892
from_path, from_entry = from_data.get(file_id, (None, None))
893
from_versioned = (from_entry is not None)
894
if from_entry is not None:
895
from_versioned = True
896
from_name = from_entry.name
897
from_parent = from_entry.parent_id
898
from_kind, from_executable, from_stat = \
899
self.source._comparison_data(from_entry, from_path)
902
from_versioned = False
906
from_executable = None
907
versioned = (from_versioned, True)
908
to_kind, to_executable, to_stat = \
909
self.target._comparison_data(to_entry, to_path)
910
kind = (from_kind, to_kind)
911
if kind[0] != kind[1]:
912
changed_content = True
913
elif from_kind == 'file':
914
from_size = self.source._file_size(from_entry, from_stat)
915
to_size = self.target._file_size(to_entry, to_stat)
916
if from_size != to_size:
917
changed_content = True
918
elif (self.source.get_file_sha1(file_id, from_path, from_stat) !=
919
self.target.get_file_sha1(file_id, to_path, to_stat)):
920
changed_content = True
921
elif from_kind == 'symlink':
922
if (self.source.get_symlink_target(file_id) !=
923
self.target.get_symlink_target(file_id)):
924
changed_content = True
925
elif from_kind == 'tree-reference':
926
if (self.source.get_reference_revision(file_id, from_path)
927
!= self.target.get_reference_revision(file_id, to_path)):
928
changed_content = True
929
parent = (from_parent, to_entry.parent_id)
930
name = (from_name, to_entry.name)
931
executable = (from_executable, to_executable)
933
pb.update('comparing files', entry_count, num_entries)
934
if (changed_content is not False or versioned[0] != versioned[1]
935
or parent[0] != parent[1] or name[0] != name[1] or
936
executable[0] != executable[1] or include_unchanged):
937
yield (file_id, (from_path, to_path), changed_content,
938
versioned, parent, name, kind, executable)
940
while all_unversioned:
941
# yield any trailing unversioned paths
942
unversioned_path = all_unversioned.popleft()
943
to_kind, to_executable, to_stat = \
944
self.target._comparison_data(fake_entry, unversioned_path[1])
945
yield (None, (None, unversioned_path[1]), True, (False, False),
947
(None, unversioned_path[0][-1]),
949
(None, to_executable))
951
def get_to_path(to_entry):
952
if to_entry.parent_id is None:
953
to_path = '' # the root
955
if to_entry.parent_id not in to_paths:
957
return get_to_path(self.target.inventory[to_entry.parent_id])
958
to_path = osutils.pathjoin(to_paths[to_entry.parent_id],
960
to_paths[to_entry.file_id] = to_path
963
for path, from_entry in from_entries_by_dir:
964
file_id = from_entry.file_id
965
if file_id in to_paths:
968
if not file_id in self.target.all_file_ids():
969
# common case - paths we have not emitted are not present in
973
to_path = get_to_path(self.target.inventory[file_id])
976
pb.update('comparing files', entry_count, num_entries)
977
versioned = (True, False)
978
parent = (from_entry.parent_id, None)
979
name = (from_entry.name, None)
980
from_kind, from_executable, stat_value = \
981
self.source._comparison_data(from_entry, path)
982
kind = (from_kind, None)
983
executable = (from_executable, None)
984
changed_content = from_kind is not None
985
# the parent's path is necessarily known at this point.
986
yield(file_id, (path, to_path), changed_content, versioned, parent,
987
name, kind, executable)
990
class MultiWalker(object):
991
"""Walk multiple trees simultaneously, getting combined results."""
993
# Note: This could be written to not assume you can do out-of-order
994
# lookups. Instead any nodes that don't match in all trees could be
995
# marked as 'deferred', and then returned in the final cleanup loop.
996
# For now, I think it is "nicer" to return things as close to the
997
# "master_tree" order as we can.
999
def __init__(self, master_tree, other_trees):
1000
"""Create a new MultiWalker.
1002
All trees being walked must implement "iter_entries_by_dir()", such
1003
that they yield (path, object) tuples, where that object will have a
1004
'.file_id' member, that can be used to check equality.
1006
:param master_tree: All trees will be 'slaved' to the master_tree such
1007
that nodes in master_tree will be used as 'first-pass' sync points.
1008
Any nodes that aren't in master_tree will be merged in a second
1010
:param other_trees: A list of other trees to walk simultaneously.
1012
self._master_tree = master_tree
1013
self._other_trees = other_trees
1015
# Keep track of any nodes that were properly processed just out of
1016
# order, that way we don't return them at the end, we don't have to
1017
# track *all* processed file_ids, just the out-of-order ones
1018
self._out_of_order_processed = set()
1021
def _step_one(iterator):
1022
"""Step an iter_entries_by_dir iterator.
1024
:return: (has_more, path, ie)
1025
If has_more is False, path and ie will be None.
1028
path, ie = iterator.next()
1029
except StopIteration:
1030
return False, None, None
1032
return True, path, ie
1035
def _cmp_path_by_dirblock(path1, path2):
1036
"""Compare two paths based on what directory they are in.
1038
This generates a sort order, such that all children of a directory are
1039
sorted together, and grandchildren are in the same order as the
1040
children appear. But all grandchildren come after all children.
1042
:param path1: first path
1043
:param path2: the second path
1044
:return: negative number if ``path1`` comes first,
1045
0 if paths are equal
1046
and a positive number if ``path2`` sorts first
1048
# Shortcut this special case
1051
# This is stolen from _dirstate_helpers_py.py, only switching it to
1052
# Unicode objects. Consider using encode_utf8() and then using the
1053
# optimized versions, or maybe writing optimized unicode versions.
1054
if not isinstance(path1, unicode):
1055
raise TypeError("'path1' must be a unicode string, not %s: %r"
1056
% (type(path1), path1))
1057
if not isinstance(path2, unicode):
1058
raise TypeError("'path2' must be a unicode string, not %s: %r"
1059
% (type(path2), path2))
1060
return cmp(MultiWalker._path_to_key(path1),
1061
MultiWalker._path_to_key(path2))
1064
def _path_to_key(path):
1065
dirname, basename = osutils.split(path)
1066
return (dirname.split(u'/'), basename)
1068
def _lookup_by_file_id(self, extra_entries, other_tree, file_id):
1069
"""Lookup an inventory entry by file_id.
1071
This is called when an entry is missing in the normal order.
1072
Generally this is because a file was either renamed, or it was
1073
deleted/added. If the entry was found in the inventory and not in
1074
extra_entries, it will be added to self._out_of_order_processed
1076
:param extra_entries: A dictionary of {file_id: (path, ie)}. This
1077
should be filled with entries that were found before they were
1078
used. If file_id is present, it will be removed from the
1080
:param other_tree: The Tree to search, in case we didn't find the entry
1082
:param file_id: The file_id to look for
1083
:return: (path, ie) if found or (None, None) if not present.
1085
if file_id in extra_entries:
1086
return extra_entries.pop(file_id)
1087
# TODO: Is id2path better as the first call, or is
1088
# inventory[file_id] better as a first check?
1090
cur_path = other_tree.id2path(file_id)
1091
except errors.NoSuchId:
1093
if cur_path is None:
1096
self._out_of_order_processed.add(file_id)
1097
cur_ie = other_tree.inventory[file_id]
1098
return (cur_path, cur_ie)
1101
"""Match up the values in the different trees."""
1102
for result in self._walk_master_tree():
1104
self._finish_others()
1105
for result in self._walk_others():
1108
def _walk_master_tree(self):
1109
"""First pass, walk all trees in lock-step.
1111
When we are done, all nodes in the master_tree will have been
1112
processed. _other_walkers, _other_entries, and _others_extra will be
1113
set on 'self' for future processing.
1115
# This iterator has the most "inlining" done, because it tends to touch
1116
# every file in the tree, while the others only hit nodes that don't
1118
master_iterator = self._master_tree.iter_entries_by_dir()
1120
other_walkers = [other.iter_entries_by_dir()
1121
for other in self._other_trees]
1122
other_entries = [self._step_one(walker) for walker in other_walkers]
1123
# Track extra nodes in the other trees
1124
others_extra = [{} for i in xrange(len(self._other_trees))]
1126
master_has_more = True
1127
step_one = self._step_one
1128
lookup_by_file_id = self._lookup_by_file_id
1129
out_of_order_processed = self._out_of_order_processed
1131
while master_has_more:
1132
(master_has_more, path, master_ie) = step_one(master_iterator)
1133
if not master_has_more:
1136
file_id = master_ie.file_id
1138
other_values_append = other_values.append
1139
next_other_entries = []
1140
next_other_entries_append = next_other_entries.append
1141
for idx, (other_has_more, other_path, other_ie) in enumerate(other_entries):
1142
if not other_has_more:
1143
other_values_append(lookup_by_file_id(
1144
others_extra[idx], self._other_trees[idx], file_id))
1145
next_other_entries_append((False, None, None))
1146
elif file_id == other_ie.file_id:
1147
# This is the critical code path, as most of the entries
1148
# should match between most trees.
1149
other_values_append((other_path, other_ie))
1150
next_other_entries_append(step_one(other_walkers[idx]))
1152
# This walker did not match, step it until it either
1153
# matches, or we know we are past the current walker.
1154
other_walker = other_walkers[idx]
1155
other_extra = others_extra[idx]
1156
while (other_has_more and
1157
self._cmp_path_by_dirblock(other_path, path) < 0):
1158
other_file_id = other_ie.file_id
1159
if other_file_id not in out_of_order_processed:
1160
other_extra[other_file_id] = (other_path, other_ie)
1161
other_has_more, other_path, other_ie = \
1162
step_one(other_walker)
1163
if other_has_more and other_ie.file_id == file_id:
1164
# We ended up walking to this point, match and step
1166
other_values_append((other_path, other_ie))
1167
other_has_more, other_path, other_ie = \
1168
step_one(other_walker)
1170
# This record isn't in the normal order, see if it
1172
other_values_append(lookup_by_file_id(
1173
other_extra, self._other_trees[idx], file_id))
1174
next_other_entries_append((other_has_more, other_path,
1176
other_entries = next_other_entries
1178
# We've matched all the walkers, yield this datapoint
1179
yield path, file_id, master_ie, other_values
1180
self._other_walkers = other_walkers
1181
self._other_entries = other_entries
1182
self._others_extra = others_extra
1184
def _finish_others(self):
1185
"""Finish walking the other iterators, so we get all entries."""
1186
for idx, info in enumerate(self._other_entries):
1187
other_extra = self._others_extra[idx]
1188
(other_has_more, other_path, other_ie) = info
1189
while other_has_more:
1190
other_file_id = other_ie.file_id
1191
if other_file_id not in self._out_of_order_processed:
1192
other_extra[other_file_id] = (other_path, other_ie)
1193
other_has_more, other_path, other_ie = \
1194
self._step_one(self._other_walkers[idx])
1195
del self._other_entries
1197
def _walk_others(self):
1198
"""Finish up by walking all the 'deferred' nodes."""
1199
# TODO: One alternative would be to grab all possible unprocessed
1200
# file_ids, and then sort by path, and then yield them. That
1201
# might ensure better ordering, in case a caller strictly
1202
# requires parents before children.
1203
for idx, other_extra in enumerate(self._others_extra):
1204
others = sorted(other_extra.itervalues(),
1205
key=lambda x: self._path_to_key(x[0]))
1206
for other_path, other_ie in others:
1207
file_id = other_ie.file_id
1208
# We don't need to check out_of_order_processed here, because
1209
# the lookup_by_file_id will be removing anything processed
1210
# from the extras cache
1211
other_extra.pop(file_id)
1212
other_values = [(None, None) for i in xrange(idx)]
1213
other_values.append((other_path, other_ie))
1214
for alt_idx, alt_extra in enumerate(self._others_extra[idx+1:]):
1215
alt_idx = alt_idx + idx + 1
1216
alt_extra = self._others_extra[alt_idx]
1217
alt_tree = self._other_trees[alt_idx]
1218
other_values.append(self._lookup_by_file_id(
1219
alt_extra, alt_tree, file_id))
1220
yield other_path, file_id, None, other_values
added
removed
bzrlib/tree.py
