1
# Copyright (C) 2005-2011 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., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
17
"""Tree classes, representing directory at point in time.
22
from bzrlib.lazy_import import lazy_import
23
lazy_import(globals(), """
27
conflicts as _mod_conflicts,
34
revision as _mod_revision,
40
from bzrlib.decorators import needs_read_lock
41
from bzrlib.inter import InterObject
42
from bzrlib.symbol_versioning import (
49
"""Abstract file tree.
51
There are several subclasses:
53
* `WorkingTree` exists as files on disk editable by the user.
55
* `RevisionTree` is a tree as recorded at some point in the past.
57
Trees can be compared, etc, regardless of whether they are working
58
trees or versioned trees.
61
def changes_from(self, other, want_unchanged=False, specific_files=None,
62
extra_trees=None, require_versioned=False, include_root=False,
63
want_unversioned=False):
64
"""Return a TreeDelta of the changes from other to this tree.
66
:param other: A tree to compare with.
67
:param specific_files: An optional list of file paths to restrict the
68
comparison to. When mapping filenames to ids, all matches in all
69
trees (including optional extra_trees) are used, and all children of
70
matched directories are included.
71
:param want_unchanged: An optional boolean requesting the inclusion of
72
unchanged entries in the result.
73
:param extra_trees: An optional list of additional trees to use when
74
mapping the contents of specific_files (paths) to file_ids.
75
:param require_versioned: An optional boolean (defaults to False). When
76
supplied and True all the 'specific_files' must be versioned, or
77
a PathsNotVersionedError will be thrown.
78
:param want_unversioned: Scan for unversioned paths.
80
The comparison will be performed by an InterTree object looked up on
83
# Martin observes that Tree.changes_from returns a TreeDelta and this
84
# may confuse people, because the class name of the returned object is
85
# a synonym of the object referenced in the method name.
86
return InterTree.get(other, self).compare(
87
want_unchanged=want_unchanged,
88
specific_files=specific_files,
89
extra_trees=extra_trees,
90
require_versioned=require_versioned,
91
include_root=include_root,
92
want_unversioned=want_unversioned,
95
def iter_changes(self, from_tree, include_unchanged=False,
96
specific_files=None, pb=None, extra_trees=None,
97
require_versioned=True, want_unversioned=False):
98
"""See InterTree.iter_changes"""
99
intertree = InterTree.get(from_tree, self)
100
return intertree.iter_changes(include_unchanged, specific_files, pb,
101
extra_trees, require_versioned, want_unversioned=want_unversioned)
104
"""Get a list of the conflicts in the tree.
106
Each conflict is an instance of bzrlib.conflicts.Conflict.
108
return _mod_conflicts.ConflictList()
111
"""For trees that can have unversioned files, return all such paths."""
114
def get_parent_ids(self):
115
"""Get the parent ids for this tree.
117
:return: a list of parent ids. [] is returned to indicate
118
a tree with no parents.
119
:raises: BzrError if the parents are not known.
121
raise NotImplementedError(self.get_parent_ids)
123
def has_filename(self, filename):
124
"""True if the tree has given filename."""
125
raise NotImplementedError(self.has_filename)
127
def has_id(self, file_id):
128
raise NotImplementedError(self.has_id)
130
def __contains__(self, file_id):
131
return self.has_id(file_id)
133
def has_or_had_id(self, file_id):
134
raise NotImplementedError(self.has_or_had_id)
136
def is_ignored(self, filename):
137
"""Check whether the filename is ignored by this tree.
139
:param filename: The relative filename within the tree.
140
:return: True if the filename is ignored.
144
def all_file_ids(self):
145
"""Iterate through all file ids, including ids for missing files."""
146
raise NotImplementedError(self.all_file_ids)
148
def id2path(self, file_id):
149
"""Return the path for a file id.
153
raise NotImplementedError(self.id2path)
155
def iter_entries_by_dir(self, specific_file_ids=None, yield_parents=False):
156
"""Walk the tree in 'by_dir' order.
158
This will yield each entry in the tree as a (path, entry) tuple.
159
The order that they are yielded is:
161
Directories are walked in a depth-first lexicographical order,
162
however, whenever a directory is reached, all of its direct child
163
nodes are yielded in lexicographical order before yielding the
166
For example, in the tree::
176
The yield order (ignoring root) would be::
177
a, f, a/b, a/d, a/b/c, a/d/e, f/g
179
:param yield_parents: If True, yield the parents from the root leading
180
down to specific_file_ids that have been requested. This has no
181
impact if specific_file_ids is None.
183
raise NotImplementedError(self.iter_entries_by_dir)
185
def list_files(self, include_root=False, from_dir=None, recursive=True):
186
"""List all files in this tree.
188
:param include_root: Whether to include the entry for the tree root
189
:param from_dir: Directory under which to list files
190
:param recursive: Whether to list files recursively
191
:return: iterator over tuples of (path, versioned, kind, file_id,
194
raise NotImplementedError(self.list_files)
196
def iter_references(self):
197
if self.supports_tree_reference():
198
for path, entry in self.iter_entries_by_dir():
199
if entry.kind == 'tree-reference':
200
yield path, entry.file_id
202
def kind(self, file_id):
203
raise NotImplementedError("Tree subclass %s must implement kind"
204
% self.__class__.__name__)
206
def stored_kind(self, file_id):
207
"""File kind stored for this file_id.
209
May not match kind on disk for working trees. Always available
210
for versioned files, even when the file itself is missing.
212
return self.kind(file_id)
214
def path_content_summary(self, path):
215
"""Get a summary of the information about path.
217
All the attributes returned are for the canonical form, not the
218
convenient form (if content filters are in use.)
220
:param path: A relative path within the tree.
221
:return: A tuple containing kind, size, exec, sha1-or-link.
222
Kind is always present (see tree.kind()).
223
size is present if kind is file and the size of the
224
canonical form can be cheaply determined, None otherwise.
225
exec is None unless kind is file and the platform supports the 'x'
227
sha1-or-link is the link target if kind is symlink, or the sha1 if
228
it can be obtained without reading the file.
230
raise NotImplementedError(self.path_content_summary)
232
def get_reference_revision(self, file_id, path=None):
233
raise NotImplementedError("Tree subclass %s must implement "
234
"get_reference_revision"
235
% self.__class__.__name__)
237
def _comparison_data(self, entry, path):
238
"""Return a tuple of kind, executable, stat_value for a file.
240
entry may be None if there is no inventory entry for the file, but
241
path must always be supplied.
243
kind is None if there is no file present (even if an inventory id is
244
present). executable is False for non-file entries.
246
raise NotImplementedError(self._comparison_data)
248
def _file_size(self, entry, stat_value):
249
raise NotImplementedError(self._file_size)
251
def get_file(self, file_id, path=None):
252
"""Return a file object for the file file_id in the tree.
254
If both file_id and path are defined, it is implementation defined as
255
to which one is used.
257
raise NotImplementedError(self.get_file)
259
def get_file_with_stat(self, file_id, path=None):
260
"""Get a file handle and stat object for file_id.
262
The default implementation returns (self.get_file, None) for backwards
265
:param file_id: The file id to read.
266
:param path: The path of the file, if it is known.
267
:return: A tuple (file_handle, stat_value_or_None). If the tree has
268
no stat facility, or need for a stat cache feedback during commit,
269
it may return None for the second element of the tuple.
271
return (self.get_file(file_id, path), None)
273
def get_file_text(self, file_id, path=None):
274
"""Return the byte content of a file.
276
:param file_id: The file_id of the file.
277
:param path: The path of the file.
278
If both file_id and path are supplied, an implementation may use
281
my_file = self.get_file(file_id, path)
283
return my_file.read()
287
def get_file_lines(self, file_id, path=None):
288
"""Return the content of a file, as lines.
290
:param file_id: The file_id of the file.
291
:param path: The path of the file.
292
If both file_id and path are supplied, an implementation may use
295
return osutils.split_lines(self.get_file_text(file_id, path))
297
def get_file_sha1(self, file_id, path=None):
298
"""Return the SHA1 file for a file.
300
:param file_id: The handle for this file.
301
:param path: The path that this file can be found at.
302
These must point to the same object.
304
raise NotImplementedError(self.get_file_sha1)
306
def get_file_mtime(self, file_id, path=None):
307
"""Return the modification time for a file.
309
:param file_id: The handle for this file.
310
:param path: The path that this file can be found at.
311
These must point to the same object.
313
raise NotImplementedError(self.get_file_mtime)
315
def get_file_size(self, file_id):
316
"""Return the size of a file in bytes.
318
This applies only to regular files. If invoked on directories or
319
symlinks, it will return None.
320
:param file_id: The file-id of the file
322
raise NotImplementedError(self.get_file_size)
324
def get_file_by_path(self, path):
325
raise NotImplementedError(self.get_file_by_path)
327
def is_executable(self, file_id, path=None):
328
"""Check if a file is executable.
330
:param file_id: The handle for this file.
331
:param path: The path that this file can be found at.
332
These must point to the same object.
334
raise NotImplementedError(self.is_executable)
336
def iter_files_bytes(self, desired_files):
337
"""Iterate through file contents.
339
Files will not necessarily be returned in the order they occur in
340
desired_files. No specific order is guaranteed.
342
Yields pairs of identifier, bytes_iterator. identifier is an opaque
343
value supplied by the caller as part of desired_files. It should
344
uniquely identify the file version in the caller's context. (Examples:
345
an index number or a TreeTransform trans_id.)
347
bytes_iterator is an iterable of bytestrings for the file. The
348
kind of iterable and length of the bytestrings are unspecified, but for
349
this implementation, it is a tuple containing a single bytestring with
350
the complete text of the file.
352
:param desired_files: a list of (file_id, identifier) pairs
354
for file_id, identifier in desired_files:
355
# We wrap the string in a tuple so that we can return an iterable
356
# of bytestrings. (Technically, a bytestring is also an iterable
357
# of bytestrings, but iterating through each character is not
359
cur_file = (self.get_file_text(file_id),)
360
yield identifier, cur_file
362
def get_symlink_target(self, file_id, path=None):
363
"""Get the target for a given file_id.
365
It is assumed that the caller already knows that file_id is referencing
367
:param file_id: Handle for the symlink entry.
368
:param path: The path of the file.
369
If both file_id and path are supplied, an implementation may use
371
:return: The path the symlink points to.
373
raise NotImplementedError(self.get_symlink_target)
375
def get_root_id(self):
376
"""Return the file_id for the root of this tree."""
377
raise NotImplementedError(self.get_root_id)
379
def annotate_iter(self, file_id,
380
default_revision=_mod_revision.CURRENT_REVISION):
381
"""Return an iterator of revision_id, line tuples.
383
For working trees (and mutable trees in general), the special
384
revision_id 'current:' will be used for lines that are new in this
385
tree, e.g. uncommitted changes.
386
:param file_id: The file to produce an annotated version from
387
:param default_revision: For lines that don't match a basis, mark them
388
with this revision id. Not all implementations will make use of
391
raise NotImplementedError(self.annotate_iter)
393
def _get_plan_merge_data(self, file_id, other, base):
394
from bzrlib import versionedfile
395
vf = versionedfile._PlanMergeVersionedFile(file_id)
396
last_revision_a = self._get_file_revision(file_id, vf, 'this:')
397
last_revision_b = other._get_file_revision(file_id, vf, 'other:')
399
last_revision_base = None
401
last_revision_base = base._get_file_revision(file_id, vf, 'base:')
402
return vf, last_revision_a, last_revision_b, last_revision_base
404
def plan_file_merge(self, file_id, other, base=None):
405
"""Generate a merge plan based on annotations.
407
If the file contains uncommitted changes in this tree, they will be
408
attributed to the 'current:' pseudo-revision. If the file contains
409
uncommitted changes in the other tree, they will be assigned to the
410
'other:' pseudo-revision.
412
data = self._get_plan_merge_data(file_id, other, base)
413
vf, last_revision_a, last_revision_b, last_revision_base = data
414
return vf.plan_merge(last_revision_a, last_revision_b,
417
def plan_file_lca_merge(self, file_id, other, base=None):
418
"""Generate a merge plan based lca-newness.
420
If the file contains uncommitted changes in this tree, they will be
421
attributed to the 'current:' pseudo-revision. If the file contains
422
uncommitted changes in the other tree, they will be assigned to the
423
'other:' pseudo-revision.
425
data = self._get_plan_merge_data(file_id, other, base)
426
vf, last_revision_a, last_revision_b, last_revision_base = data
427
return vf.plan_lca_merge(last_revision_a, last_revision_b,
430
def _iter_parent_trees(self):
431
"""Iterate through parent trees, defaulting to Tree.revision_tree."""
432
for revision_id in self.get_parent_ids():
434
yield self.revision_tree(revision_id)
435
except errors.NoSuchRevisionInTree:
436
yield self.repository.revision_tree(revision_id)
438
def _get_file_revision(self, file_id, vf, tree_revision):
439
"""Ensure that file_id, tree_revision is in vf to plan the merge."""
441
if getattr(self, '_repository', None) is None:
442
last_revision = tree_revision
443
parent_keys = [(file_id, t.get_file_revision(file_id)) for t in
444
self._iter_parent_trees()]
445
vf.add_lines((file_id, last_revision), parent_keys,
446
self.get_file_lines(file_id))
447
repo = self.branch.repository
450
last_revision = self.get_file_revision(file_id)
451
base_vf = self._repository.texts
452
if base_vf not in vf.fallback_versionedfiles:
453
vf.fallback_versionedfiles.append(base_vf)
456
def _check_retrieved(self, ie, f):
459
fp = osutils.fingerprint_file(f)
462
if ie.text_size is not None:
463
if ie.text_size != fp['size']:
464
raise errors.BzrError(
465
"mismatched size for file %r in %r" %
466
(ie.file_id, self._store),
467
["inventory expects %d bytes" % ie.text_size,
468
"file is actually %d bytes" % fp['size'],
469
"store is probably damaged/corrupt"])
471
if ie.text_sha1 != fp['sha1']:
472
raise errors.BzrError("wrong SHA-1 for file %r in %r" %
473
(ie.file_id, self._store),
474
["inventory expects %s" % ie.text_sha1,
475
"file is actually %s" % fp['sha1'],
476
"store is probably damaged/corrupt"])
478
def path2id(self, path):
479
"""Return the id for path in this tree."""
480
raise NotImplementedError(self.path2id)
482
def paths2ids(self, paths, trees=[], require_versioned=True):
483
"""Return all the ids that can be reached by walking from paths.
485
Each path is looked up in this tree and any extras provided in
486
trees, and this is repeated recursively: the children in an extra tree
487
of a directory that has been renamed under a provided path in this tree
488
are all returned, even if none exist under a provided path in this
489
tree, and vice versa.
491
:param paths: An iterable of paths to start converting to ids from.
492
Alternatively, if paths is None, no ids should be calculated and None
493
will be returned. This is offered to make calling the api unconditional
494
for code that *might* take a list of files.
495
:param trees: Additional trees to consider.
496
:param require_versioned: If False, do not raise NotVersionedError if
497
an element of paths is not versioned in this tree and all of trees.
499
return find_ids_across_trees(paths, [self] + list(trees), require_versioned)
501
def iter_children(self, file_id):
502
entry = self.iter_entries_by_dir([file_id]).next()[1]
503
for child in getattr(entry, 'children', {}).itervalues():
507
"""Lock this tree for multiple read only operations.
509
:return: A bzrlib.lock.LogicalLockResult.
513
def revision_tree(self, revision_id):
514
"""Obtain a revision tree for the revision revision_id.
516
The intention of this method is to allow access to possibly cached
517
tree data. Implementors of this method should raise NoSuchRevision if
518
the tree is not locally available, even if they could obtain the
519
tree via a repository or some other means. Callers are responsible
520
for finding the ultimate source for a revision tree.
522
:param revision_id: The revision_id of the requested tree.
524
:raises: NoSuchRevision if the tree cannot be obtained.
526
raise errors.NoSuchRevisionInTree(self, revision_id)
529
"""What files are present in this tree and unknown.
531
:return: an iterator over the unknown files.
538
def filter_unversioned_files(self, paths):
539
"""Filter out paths that are versioned.
541
:return: set of paths.
543
raise NotImplementedError(self.filter_unversioned_files)
545
def walkdirs(self, prefix=""):
546
"""Walk the contents of this tree from path down.
548
This yields all the data about the contents of a directory at a time.
549
After each directory has been yielded, if the caller has mutated the
550
list to exclude some directories, they are then not descended into.
552
The data yielded is of the form:
553
((directory-relpath, directory-path-from-root, directory-fileid),
554
[(relpath, basename, kind, lstat, path_from_tree_root, file_id,
555
versioned_kind), ...]),
556
- directory-relpath is the containing dirs relpath from prefix
557
- directory-path-from-root is the containing dirs path from /
558
- directory-fileid is the id of the directory if it is versioned.
559
- relpath is the relative path within the subtree being walked.
560
- basename is the basename
561
- kind is the kind of the file now. If unknonwn then the file is not
562
present within the tree - but it may be recorded as versioned. See
564
- lstat is the stat data *if* the file was statted.
565
- path_from_tree_root is the path from the root of the tree.
566
- file_id is the file_id if the entry is versioned.
567
- versioned_kind is the kind of the file as last recorded in the
568
versioning system. If 'unknown' the file is not versioned.
569
One of 'kind' and 'versioned_kind' must not be 'unknown'.
571
:param prefix: Start walking from prefix within the tree rather than
572
at the root. This allows one to walk a subtree but get paths that are
573
relative to a tree rooted higher up.
574
:return: an iterator over the directory data.
576
raise NotImplementedError(self.walkdirs)
578
def supports_content_filtering(self):
581
def _content_filter_stack(self, path=None, file_id=None):
582
"""The stack of content filters for a path if filtering is supported.
584
Readers will be applied in first-to-last order.
585
Writers will be applied in last-to-first order.
586
Either the path or the file-id needs to be provided.
588
:param path: path relative to the root of the tree
590
:param file_id: file_id or None if unknown
591
:return: the list of filters - [] if there are none
593
filter_pref_names = filters._get_registered_names()
594
if len(filter_pref_names) == 0:
597
path = self.id2path(file_id)
598
prefs = self.iter_search_rules([path], filter_pref_names).next()
599
stk = filters._get_filter_stack_for(prefs)
600
if 'filters' in debug.debug_flags:
601
trace.note("*** %s content-filter: %s => %r" % (path,prefs,stk))
604
def _content_filter_stack_provider(self):
605
"""A function that returns a stack of ContentFilters.
607
The function takes a path (relative to the top of the tree) and a
608
file-id as parameters.
610
:return: None if content filtering is not supported by this tree.
612
if self.supports_content_filtering():
613
return lambda path, file_id: \
614
self._content_filter_stack(path, file_id)
618
def iter_search_rules(self, path_names, pref_names=None,
619
_default_searcher=None):
620
"""Find the preferences for filenames in a tree.
622
:param path_names: an iterable of paths to find attributes for.
623
Paths are given relative to the root of the tree.
624
:param pref_names: the list of preferences to lookup - None for all
625
:param _default_searcher: private parameter to assist testing - don't use
626
:return: an iterator of tuple sequences, one per path-name.
627
See _RulesSearcher.get_items for details on the tuple sequence.
629
if _default_searcher is None:
630
_default_searcher = rules._per_user_searcher
631
searcher = self._get_rules_searcher(_default_searcher)
632
if searcher is not None:
633
if pref_names is not None:
634
for path in path_names:
635
yield searcher.get_selected_items(path, pref_names)
637
for path in path_names:
638
yield searcher.get_items(path)
640
def _get_rules_searcher(self, default_searcher):
641
"""Get the RulesSearcher for this tree given the default one."""
642
searcher = default_searcher
646
class InventoryTree(Tree):
647
"""A tree that relies on an inventory for its metadata.
649
Trees contain an `Inventory` object, and also know how to retrieve
650
file texts mentioned in the inventory, either from a working
651
directory or from a store.
653
It is possible for trees to contain files that are not described
654
in their inventory or vice versa; for this use `filenames()`.
656
Subclasses should set the _inventory attribute, which is considered
657
private to external API users.
660
def get_canonical_inventory_paths(self, paths):
661
"""Like get_canonical_inventory_path() but works on multiple items.
663
:param paths: A sequence of paths relative to the root of the tree.
664
:return: A list of paths, with each item the corresponding input path
665
adjusted to account for existing elements that match case
668
return list(self._yield_canonical_inventory_paths(paths))
670
def get_canonical_inventory_path(self, path):
671
"""Returns the first inventory item that case-insensitively matches path.
673
If a path matches exactly, it is returned. If no path matches exactly
674
but more than one path matches case-insensitively, it is implementation
675
defined which is returned.
677
If no path matches case-insensitively, the input path is returned, but
678
with as many path entries that do exist changed to their canonical
681
If you need to resolve many names from the same tree, you should
682
use get_canonical_inventory_paths() to avoid O(N) behaviour.
684
:param path: A paths relative to the root of the tree.
685
:return: The input path adjusted to account for existing elements
686
that match case insensitively.
688
return self._yield_canonical_inventory_paths([path]).next()
690
def _yield_canonical_inventory_paths(self, paths):
692
# First, if the path as specified exists exactly, just use it.
693
if self.path2id(path) is not None:
697
cur_id = self.get_root_id()
699
bit_iter = iter(path.split("/"))
703
for child in self.iter_children(cur_id):
705
# XXX: it seem like if the child is known to be in the
706
# tree, we shouldn't need to go from its id back to
707
# its path -- mbp 2010-02-11
709
# XXX: it seems like we could be more efficient
710
# by just directly looking up the original name and
711
# only then searching all children; also by not
712
# chopping paths so much. -- mbp 2010-02-11
713
child_base = os.path.basename(self.id2path(child))
714
if (child_base == elt):
715
# if we found an exact match, we can stop now; if
716
# we found an approximate match we need to keep
717
# searching because there might be an exact match
720
new_path = osutils.pathjoin(cur_path, child_base)
722
elif child_base.lower() == lelt:
724
new_path = osutils.pathjoin(cur_path, child_base)
725
except errors.NoSuchId:
726
# before a change is committed we can see this error...
731
# got to the end of this directory and no entries matched.
732
# Return what matched so far, plus the rest as specified.
733
cur_path = osutils.pathjoin(cur_path, elt, *list(bit_iter))
738
def _get_inventory(self):
739
return self._inventory
741
inventory = property(_get_inventory,
742
doc="Inventory of this Tree")
745
def path2id(self, path):
746
"""Return the id for path in this tree."""
747
return self._inventory.path2id(path)
749
def id2path(self, file_id):
750
"""Return the path for a file id.
754
return self.inventory.id2path(file_id)
756
def has_id(self, file_id):
757
return self.inventory.has_id(file_id)
759
def has_or_had_id(self, file_id):
760
return self.inventory.has_id(file_id)
762
def all_file_ids(self):
763
return set(self.inventory)
765
@deprecated_method(deprecated_in((2, 4, 0)))
767
return iter(self.inventory)
769
def filter_unversioned_files(self, paths):
770
"""Filter out paths that are versioned.
772
:return: set of paths.
774
# NB: we specifically *don't* call self.has_filename, because for
775
# WorkingTrees that can indicate files that exist on disk but that
777
pred = self.inventory.has_filename
778
return set((p for p in paths if not pred(p)))
781
def iter_entries_by_dir(self, specific_file_ids=None, yield_parents=False):
782
"""Walk the tree in 'by_dir' order.
784
This will yield each entry in the tree as a (path, entry) tuple.
785
The order that they are yielded is:
787
See Tree.iter_entries_by_dir for details.
789
:param yield_parents: If True, yield the parents from the root leading
790
down to specific_file_ids that have been requested. This has no
791
impact if specific_file_ids is None.
793
return self.inventory.iter_entries_by_dir(
794
specific_file_ids=specific_file_ids, yield_parents=yield_parents)
796
def get_file_by_path(self, path):
797
return self.get_file(self._inventory.path2id(path), path)
800
######################################################################
803
# TODO: Merge these two functions into a single one that can operate
804
# on either a whole tree or a set of files.
806
# TODO: Return the diff in order by filename, not by category or in
807
# random order. Can probably be done by lock-stepping through the
808
# filenames from both trees.
811
def file_status(filename, old_tree, new_tree):
812
"""Return single-letter status, old and new names for a file.
814
The complexity here is in deciding how to represent renames;
815
many complex cases are possible.
817
old_inv = old_tree.inventory
818
new_inv = new_tree.inventory
819
new_id = new_inv.path2id(filename)
820
old_id = old_inv.path2id(filename)
822
if not new_id and not old_id:
823
# easy: doesn't exist in either; not versioned at all
824
if new_tree.is_ignored(filename):
825
return 'I', None, None
827
return '?', None, None
829
# There is now a file of this name, great.
832
# There is no longer a file of this name, but we can describe
833
# what happened to the file that used to have
834
# this name. There are two possibilities: either it was
835
# deleted entirely, or renamed.
836
if new_inv.has_id(old_id):
837
return 'X', old_inv.id2path(old_id), new_inv.id2path(old_id)
839
return 'D', old_inv.id2path(old_id), None
841
# if the file_id is new in this revision, it is added
842
if new_id and not old_inv.has_id(new_id):
845
# if there used to be a file of this name, but that ID has now
846
# disappeared, it is deleted
847
if old_id and not new_inv.has_id(old_id):
853
def find_ids_across_trees(filenames, trees, require_versioned=True):
854
"""Find the ids corresponding to specified filenames.
856
All matches in all trees will be used, and all children of matched
857
directories will be used.
859
:param filenames: The filenames to find file_ids for (if None, returns
861
:param trees: The trees to find file_ids within
862
:param require_versioned: if true, all specified filenames must occur in
864
:return: a set of file ids for the specified filenames and their children.
868
specified_path_ids = _find_ids_across_trees(filenames, trees,
870
return _find_children_across_trees(specified_path_ids, trees)
873
def _find_ids_across_trees(filenames, trees, require_versioned):
874
"""Find the ids corresponding to specified filenames.
876
All matches in all trees will be used, but subdirectories are not scanned.
878
:param filenames: The filenames to find file_ids for
879
:param trees: The trees to find file_ids within
880
:param require_versioned: if true, all specified filenames must occur in
882
:return: a set of file ids for the specified filenames
885
interesting_ids = set()
886
for tree_path in filenames:
889
file_id = tree.path2id(tree_path)
890
if file_id is not None:
891
interesting_ids.add(file_id)
894
not_versioned.append(tree_path)
895
if len(not_versioned) > 0 and require_versioned:
896
raise errors.PathsNotVersionedError(not_versioned)
897
return interesting_ids
900
def _find_children_across_trees(specified_ids, trees):
901
"""Return a set including specified ids and their children.
903
All matches in all trees will be used.
905
:param trees: The trees to find file_ids within
906
:return: a set containing all specified ids and their children
908
interesting_ids = set(specified_ids)
909
pending = interesting_ids
910
# now handle children of interesting ids
911
# we loop so that we handle all children of each id in both trees
912
while len(pending) > 0:
914
for file_id in pending:
916
if not tree.has_or_had_id(file_id):
918
for child_id in tree.iter_children(file_id):
919
if child_id not in interesting_ids:
920
new_pending.add(child_id)
921
interesting_ids.update(new_pending)
922
pending = new_pending
923
return interesting_ids
926
class InterTree(InterObject):
927
"""This class represents operations taking place between two Trees.
929
Its instances have methods like 'compare' and contain references to the
930
source and target trees these operations are to be carried out on.
932
Clients of bzrlib should not need to use InterTree directly, rather they
933
should use the convenience methods on Tree such as 'Tree.compare()' which
934
will pass through to InterTree as appropriate.
937
# Formats that will be used to test this InterTree. If both are
938
# None, this InterTree will not be tested (e.g. because a complex
940
_matching_from_tree_format = None
941
_matching_to_tree_format = None
946
def is_compatible(kls, source, target):
947
# The default implementation is naive and uses the public API, so
948
# it works for all trees.
951
def _changes_from_entries(self, source_entry, target_entry,
952
source_path=None, target_path=None):
953
"""Generate a iter_changes tuple between source_entry and target_entry.
955
:param source_entry: An inventory entry from self.source, or None.
956
:param target_entry: An inventory entry from self.target, or None.
957
:param source_path: The path of source_entry, if known. If not known
958
it will be looked up.
959
:param target_path: The path of target_entry, if known. If not known
960
it will be looked up.
961
:return: A tuple, item 0 of which is an iter_changes result tuple, and
962
item 1 is True if there are any changes in the result tuple.
964
if source_entry is None:
965
if target_entry is None:
967
file_id = target_entry.file_id
969
file_id = source_entry.file_id
970
if source_entry is not None:
971
source_versioned = True
972
source_name = source_entry.name
973
source_parent = source_entry.parent_id
974
if source_path is None:
975
source_path = self.source.id2path(file_id)
976
source_kind, source_executable, source_stat = \
977
self.source._comparison_data(source_entry, source_path)
979
source_versioned = False
983
source_executable = None
984
if target_entry is not None:
985
target_versioned = True
986
target_name = target_entry.name
987
target_parent = target_entry.parent_id
988
if target_path is None:
989
target_path = self.target.id2path(file_id)
990
target_kind, target_executable, target_stat = \
991
self.target._comparison_data(target_entry, target_path)
993
target_versioned = False
997
target_executable = None
998
versioned = (source_versioned, target_versioned)
999
kind = (source_kind, target_kind)
1000
changed_content = False
1001
if source_kind != target_kind:
1002
changed_content = True
1003
elif source_kind == 'file':
1004
if (self.source.get_file_sha1(file_id, source_path, source_stat) !=
1005
self.target.get_file_sha1(file_id, target_path, target_stat)):
1006
changed_content = True
1007
elif source_kind == 'symlink':
1008
if (self.source.get_symlink_target(file_id) !=
1009
self.target.get_symlink_target(file_id)):
1010
changed_content = True
1011
# XXX: Yes, the indentation below is wrong. But fixing it broke
1012
# test_merge.TestMergerEntriesLCAOnDisk.
1013
# test_nested_tree_subtree_renamed_and_modified. We'll wait for
1014
# the fix from bzr.dev -- vila 2009026
1015
elif source_kind == 'tree-reference':
1016
if (self.source.get_reference_revision(file_id, source_path)
1017
!= self.target.get_reference_revision(file_id, target_path)):
1018
changed_content = True
1019
parent = (source_parent, target_parent)
1020
name = (source_name, target_name)
1021
executable = (source_executable, target_executable)
1022
if (changed_content is not False or versioned[0] != versioned[1]
1023
or parent[0] != parent[1] or name[0] != name[1] or
1024
executable[0] != executable[1]):
1028
return (file_id, (source_path, target_path), changed_content,
1029
versioned, parent, name, kind, executable), changes
1032
def compare(self, want_unchanged=False, specific_files=None,
1033
extra_trees=None, require_versioned=False, include_root=False,
1034
want_unversioned=False):
1035
"""Return the changes from source to target.
1037
:return: A TreeDelta.
1038
:param specific_files: An optional list of file paths to restrict the
1039
comparison to. When mapping filenames to ids, all matches in all
1040
trees (including optional extra_trees) are used, and all children of
1041
matched directories are included.
1042
:param want_unchanged: An optional boolean requesting the inclusion of
1043
unchanged entries in the result.
1044
:param extra_trees: An optional list of additional trees to use when
1045
mapping the contents of specific_files (paths) to file_ids.
1046
:param require_versioned: An optional boolean (defaults to False). When
1047
supplied and True all the 'specific_files' must be versioned, or
1048
a PathsNotVersionedError will be thrown.
1049
:param want_unversioned: Scan for unversioned paths.
1051
trees = (self.source,)
1052
if extra_trees is not None:
1053
trees = trees + tuple(extra_trees)
1054
# target is usually the newer tree:
1055
specific_file_ids = self.target.paths2ids(specific_files, trees,
1056
require_versioned=require_versioned)
1057
if specific_files and not specific_file_ids:
1058
# All files are unversioned, so just return an empty delta
1059
# _compare_trees would think we want a complete delta
1060
result = delta.TreeDelta()
1061
fake_entry = inventory.InventoryFile('unused', 'unused', 'unused')
1062
result.unversioned = [(path, None,
1063
self.target._comparison_data(fake_entry, path)[0]) for path in
1066
return delta._compare_trees(self.source, self.target, want_unchanged,
1067
specific_files, include_root, extra_trees=extra_trees,
1068
require_versioned=require_versioned,
1069
want_unversioned=want_unversioned)
1071
def iter_changes(self, include_unchanged=False,
1072
specific_files=None, pb=None, extra_trees=[],
1073
require_versioned=True, want_unversioned=False):
1074
"""Generate an iterator of changes between trees.
1076
A tuple is returned:
1077
(file_id, (path_in_source, path_in_target),
1078
changed_content, versioned, parent, name, kind,
1081
Changed_content is True if the file's content has changed. This
1082
includes changes to its kind, and to a symlink's target.
1084
versioned, parent, name, kind, executable are tuples of (from, to).
1085
If a file is missing in a tree, its kind is None.
1087
Iteration is done in parent-to-child order, relative to the target
1090
There is no guarantee that all paths are in sorted order: the
1091
requirement to expand the search due to renames may result in children
1092
that should be found early being found late in the search, after
1093
lexically later results have been returned.
1094
:param require_versioned: Raise errors.PathsNotVersionedError if a
1095
path in the specific_files list is not versioned in one of
1096
source, target or extra_trees.
1097
:param specific_files: An optional list of file paths to restrict the
1098
comparison to. When mapping filenames to ids, all matches in all
1099
trees (including optional extra_trees) are used, and all children
1100
of matched directories are included. The parents in the target tree
1101
of the specific files up to and including the root of the tree are
1102
always evaluated for changes too.
1103
:param want_unversioned: Should unversioned files be returned in the
1104
output. An unversioned file is defined as one with (False, False)
1105
for the versioned pair.
1107
lookup_trees = [self.source]
1109
lookup_trees.extend(extra_trees)
1110
# The ids of items we need to examine to insure delta consistency.
1111
precise_file_ids = set()
1112
changed_file_ids = []
1113
if specific_files == []:
1114
specific_file_ids = []
1116
specific_file_ids = self.target.paths2ids(specific_files,
1117
lookup_trees, require_versioned=require_versioned)
1118
if specific_files is not None:
1119
# reparented or added entries must have their parents included
1120
# so that valid deltas can be created. The seen_parents set
1121
# tracks the parents that we need to have.
1122
# The seen_dirs set tracks directory entries we've yielded.
1123
# After outputting version object in to_entries we set difference
1124
# the two seen sets and start checking parents.
1125
seen_parents = set()
1127
if want_unversioned:
1128
all_unversioned = sorted([(p.split('/'), p) for p in
1129
self.target.extras()
1130
if specific_files is None or
1131
osutils.is_inside_any(specific_files, p)])
1132
all_unversioned = collections.deque(all_unversioned)
1134
all_unversioned = collections.deque()
1136
from_entries_by_dir = list(self.source.iter_entries_by_dir(
1137
specific_file_ids=specific_file_ids))
1138
from_data = dict((e.file_id, (p, e)) for p, e in from_entries_by_dir)
1139
to_entries_by_dir = list(self.target.iter_entries_by_dir(
1140
specific_file_ids=specific_file_ids))
1141
num_entries = len(from_entries_by_dir) + len(to_entries_by_dir)
1143
# the unversioned path lookup only occurs on real trees - where there
1144
# can be extras. So the fake_entry is solely used to look up
1145
# executable it values when execute is not supported.
1146
fake_entry = inventory.InventoryFile('unused', 'unused', 'unused')
1147
for target_path, target_entry in to_entries_by_dir:
1148
while (all_unversioned and
1149
all_unversioned[0][0] < target_path.split('/')):
1150
unversioned_path = all_unversioned.popleft()
1151
target_kind, target_executable, target_stat = \
1152
self.target._comparison_data(fake_entry, unversioned_path[1])
1153
yield (None, (None, unversioned_path[1]), True, (False, False),
1155
(None, unversioned_path[0][-1]),
1156
(None, target_kind),
1157
(None, target_executable))
1158
source_path, source_entry = from_data.get(target_entry.file_id,
1160
result, changes = self._changes_from_entries(source_entry,
1161
target_entry, source_path=source_path, target_path=target_path)
1162
to_paths[result[0]] = result[1][1]
1167
pb.update('comparing files', entry_count, num_entries)
1168
if changes or include_unchanged:
1169
if specific_file_ids is not None:
1170
new_parent_id = result[4][1]
1171
precise_file_ids.add(new_parent_id)
1172
changed_file_ids.append(result[0])
1174
# Ensure correct behaviour for reparented/added specific files.
1175
if specific_files is not None:
1176
# Record output dirs
1177
if result[6][1] == 'directory':
1178
seen_dirs.add(result[0])
1179
# Record parents of reparented/added entries.
1180
versioned = result[3]
1182
if not versioned[0] or parents[0] != parents[1]:
1183
seen_parents.add(parents[1])
1184
while all_unversioned:
1185
# yield any trailing unversioned paths
1186
unversioned_path = all_unversioned.popleft()
1187
to_kind, to_executable, to_stat = \
1188
self.target._comparison_data(fake_entry, unversioned_path[1])
1189
yield (None, (None, unversioned_path[1]), True, (False, False),
1191
(None, unversioned_path[0][-1]),
1193
(None, to_executable))
1194
# Yield all remaining source paths
1195
for path, from_entry in from_entries_by_dir:
1196
file_id = from_entry.file_id
1197
if file_id in to_paths:
1200
if not self.target.has_id(file_id):
1201
# common case - paths we have not emitted are not present in
1205
to_path = self.target.id2path(file_id)
1208
pb.update('comparing files', entry_count, num_entries)
1209
versioned = (True, False)
1210
parent = (from_entry.parent_id, None)
1211
name = (from_entry.name, None)
1212
from_kind, from_executable, stat_value = \
1213
self.source._comparison_data(from_entry, path)
1214
kind = (from_kind, None)
1215
executable = (from_executable, None)
1216
changed_content = from_kind is not None
1217
# the parent's path is necessarily known at this point.
1218
changed_file_ids.append(file_id)
1219
yield(file_id, (path, to_path), changed_content, versioned, parent,
1220
name, kind, executable)
1221
changed_file_ids = set(changed_file_ids)
1222
if specific_file_ids is not None:
1223
for result in self._handle_precise_ids(precise_file_ids,
1227
def _get_entry(self, tree, file_id):
1228
"""Get an inventory entry from a tree, with missing entries as None.
1230
If the tree raises NotImplementedError on accessing .inventory, then
1231
this is worked around using iter_entries_by_dir on just the file id
1234
:param tree: The tree to lookup the entry in.
1235
:param file_id: The file_id to lookup.
1238
inventory = tree.inventory
1239
except NotImplementedError:
1240
# No inventory available.
1242
iterator = tree.iter_entries_by_dir(specific_file_ids=[file_id])
1243
return iterator.next()[1]
1244
except StopIteration:
1248
return inventory[file_id]
1249
except errors.NoSuchId:
1252
def _handle_precise_ids(self, precise_file_ids, changed_file_ids,
1253
discarded_changes=None):
1254
"""Fill out a partial iter_changes to be consistent.
1256
:param precise_file_ids: The file ids of parents that were seen during
1258
:param changed_file_ids: The file ids of already emitted items.
1259
:param discarded_changes: An optional dict of precalculated
1260
iter_changes items which the partial iter_changes had not output
1262
:return: A generator of iter_changes items to output.
1264
# process parents of things that had changed under the users
1265
# requested paths to prevent incorrect paths or parent ids which
1266
# aren't in the tree.
1267
while precise_file_ids:
1268
precise_file_ids.discard(None)
1269
# Don't emit file_ids twice
1270
precise_file_ids.difference_update(changed_file_ids)
1271
if not precise_file_ids:
1273
# If the there was something at a given output path in source, we
1274
# have to include the entry from source in the delta, or we would
1275
# be putting this entry into a used path.
1277
for parent_id in precise_file_ids:
1279
paths.append(self.target.id2path(parent_id))
1280
except errors.NoSuchId:
1281
# This id has been dragged in from the source by delta
1282
# expansion and isn't present in target at all: we don't
1283
# need to check for path collisions on it.
1286
old_id = self.source.path2id(path)
1287
precise_file_ids.add(old_id)
1288
precise_file_ids.discard(None)
1289
current_ids = precise_file_ids
1290
precise_file_ids = set()
1291
# We have to emit all of precise_file_ids that have been altered.
1292
# We may have to output the children of some of those ids if any
1293
# directories have stopped being directories.
1294
for file_id in current_ids:
1296
if discarded_changes:
1297
result = discarded_changes.get(file_id)
1302
old_entry = self._get_entry(self.source, file_id)
1303
new_entry = self._get_entry(self.target, file_id)
1304
result, changes = self._changes_from_entries(
1305
old_entry, new_entry)
1308
# Get this parents parent to examine.
1309
new_parent_id = result[4][1]
1310
precise_file_ids.add(new_parent_id)
1312
if (result[6][0] == 'directory' and
1313
result[6][1] != 'directory'):
1314
# This stopped being a directory, the old children have
1316
if old_entry is None:
1317
# Reusing a discarded change.
1318
old_entry = self._get_entry(self.source, file_id)
1319
for child in old_entry.children.values():
1320
precise_file_ids.add(child.file_id)
1321
changed_file_ids.add(result[0])
1325
InterTree.register_optimiser(InterTree)
1328
class MultiWalker(object):
1329
"""Walk multiple trees simultaneously, getting combined results."""
1331
# Note: This could be written to not assume you can do out-of-order
1332
# lookups. Instead any nodes that don't match in all trees could be
1333
# marked as 'deferred', and then returned in the final cleanup loop.
1334
# For now, I think it is "nicer" to return things as close to the
1335
# "master_tree" order as we can.
1337
def __init__(self, master_tree, other_trees):
1338
"""Create a new MultiWalker.
1340
All trees being walked must implement "iter_entries_by_dir()", such
1341
that they yield (path, object) tuples, where that object will have a
1342
'.file_id' member, that can be used to check equality.
1344
:param master_tree: All trees will be 'slaved' to the master_tree such
1345
that nodes in master_tree will be used as 'first-pass' sync points.
1346
Any nodes that aren't in master_tree will be merged in a second
1348
:param other_trees: A list of other trees to walk simultaneously.
1350
self._master_tree = master_tree
1351
self._other_trees = other_trees
1353
# Keep track of any nodes that were properly processed just out of
1354
# order, that way we don't return them at the end, we don't have to
1355
# track *all* processed file_ids, just the out-of-order ones
1356
self._out_of_order_processed = set()
1359
def _step_one(iterator):
1360
"""Step an iter_entries_by_dir iterator.
1362
:return: (has_more, path, ie)
1363
If has_more is False, path and ie will be None.
1366
path, ie = iterator.next()
1367
except StopIteration:
1368
return False, None, None
1370
return True, path, ie
1373
def _cmp_path_by_dirblock(path1, path2):
1374
"""Compare two paths based on what directory they are in.
1376
This generates a sort order, such that all children of a directory are
1377
sorted together, and grandchildren are in the same order as the
1378
children appear. But all grandchildren come after all children.
1380
:param path1: first path
1381
:param path2: the second path
1382
:return: negative number if ``path1`` comes first,
1383
0 if paths are equal
1384
and a positive number if ``path2`` sorts first
1386
# Shortcut this special case
1389
# This is stolen from _dirstate_helpers_py.py, only switching it to
1390
# Unicode objects. Consider using encode_utf8() and then using the
1391
# optimized versions, or maybe writing optimized unicode versions.
1392
if not isinstance(path1, unicode):
1393
raise TypeError("'path1' must be a unicode string, not %s: %r"
1394
% (type(path1), path1))
1395
if not isinstance(path2, unicode):
1396
raise TypeError("'path2' must be a unicode string, not %s: %r"
1397
% (type(path2), path2))
1398
return cmp(MultiWalker._path_to_key(path1),
1399
MultiWalker._path_to_key(path2))
1402
def _path_to_key(path):
1403
dirname, basename = osutils.split(path)
1404
return (dirname.split(u'/'), basename)
1406
def _lookup_by_file_id(self, extra_entries, other_tree, file_id):
1407
"""Lookup an inventory entry by file_id.
1409
This is called when an entry is missing in the normal order.
1410
Generally this is because a file was either renamed, or it was
1411
deleted/added. If the entry was found in the inventory and not in
1412
extra_entries, it will be added to self._out_of_order_processed
1414
:param extra_entries: A dictionary of {file_id: (path, ie)}. This
1415
should be filled with entries that were found before they were
1416
used. If file_id is present, it will be removed from the
1418
:param other_tree: The Tree to search, in case we didn't find the entry
1420
:param file_id: The file_id to look for
1421
:return: (path, ie) if found or (None, None) if not present.
1423
if file_id in extra_entries:
1424
return extra_entries.pop(file_id)
1425
# TODO: Is id2path better as the first call, or is
1426
# inventory[file_id] better as a first check?
1428
cur_path = other_tree.id2path(file_id)
1429
except errors.NoSuchId:
1431
if cur_path is None:
1434
self._out_of_order_processed.add(file_id)
1435
cur_ie = other_tree.inventory[file_id]
1436
return (cur_path, cur_ie)
1439
"""Match up the values in the different trees."""
1440
for result in self._walk_master_tree():
1442
self._finish_others()
1443
for result in self._walk_others():
1446
def _walk_master_tree(self):
1447
"""First pass, walk all trees in lock-step.
1449
When we are done, all nodes in the master_tree will have been
1450
processed. _other_walkers, _other_entries, and _others_extra will be
1451
set on 'self' for future processing.
1453
# This iterator has the most "inlining" done, because it tends to touch
1454
# every file in the tree, while the others only hit nodes that don't
1456
master_iterator = self._master_tree.iter_entries_by_dir()
1458
other_walkers = [other.iter_entries_by_dir()
1459
for other in self._other_trees]
1460
other_entries = [self._step_one(walker) for walker in other_walkers]
1461
# Track extra nodes in the other trees
1462
others_extra = [{} for i in xrange(len(self._other_trees))]
1464
master_has_more = True
1465
step_one = self._step_one
1466
lookup_by_file_id = self._lookup_by_file_id
1467
out_of_order_processed = self._out_of_order_processed
1469
while master_has_more:
1470
(master_has_more, path, master_ie) = step_one(master_iterator)
1471
if not master_has_more:
1474
file_id = master_ie.file_id
1476
other_values_append = other_values.append
1477
next_other_entries = []
1478
next_other_entries_append = next_other_entries.append
1479
for idx, (other_has_more, other_path, other_ie) in enumerate(other_entries):
1480
if not other_has_more:
1481
other_values_append(lookup_by_file_id(
1482
others_extra[idx], self._other_trees[idx], file_id))
1483
next_other_entries_append((False, None, None))
1484
elif file_id == other_ie.file_id:
1485
# This is the critical code path, as most of the entries
1486
# should match between most trees.
1487
other_values_append((other_path, other_ie))
1488
next_other_entries_append(step_one(other_walkers[idx]))
1490
# This walker did not match, step it until it either
1491
# matches, or we know we are past the current walker.
1492
other_walker = other_walkers[idx]
1493
other_extra = others_extra[idx]
1494
while (other_has_more and
1495
self._cmp_path_by_dirblock(other_path, path) < 0):
1496
other_file_id = other_ie.file_id
1497
if other_file_id not in out_of_order_processed:
1498
other_extra[other_file_id] = (other_path, other_ie)
1499
other_has_more, other_path, other_ie = \
1500
step_one(other_walker)
1501
if other_has_more and other_ie.file_id == file_id:
1502
# We ended up walking to this point, match and step
1504
other_values_append((other_path, other_ie))
1505
other_has_more, other_path, other_ie = \
1506
step_one(other_walker)
1508
# This record isn't in the normal order, see if it
1510
other_values_append(lookup_by_file_id(
1511
other_extra, self._other_trees[idx], file_id))
1512
next_other_entries_append((other_has_more, other_path,
1514
other_entries = next_other_entries
1516
# We've matched all the walkers, yield this datapoint
1517
yield path, file_id, master_ie, other_values
1518
self._other_walkers = other_walkers
1519
self._other_entries = other_entries
1520
self._others_extra = others_extra
1522
def _finish_others(self):
1523
"""Finish walking the other iterators, so we get all entries."""
1524
for idx, info in enumerate(self._other_entries):
1525
other_extra = self._others_extra[idx]
1526
(other_has_more, other_path, other_ie) = info
1527
while other_has_more:
1528
other_file_id = other_ie.file_id
1529
if other_file_id not in self._out_of_order_processed:
1530
other_extra[other_file_id] = (other_path, other_ie)
1531
other_has_more, other_path, other_ie = \
1532
self._step_one(self._other_walkers[idx])
1533
del self._other_entries
1535
def _walk_others(self):
1536
"""Finish up by walking all the 'deferred' nodes."""
1537
# TODO: One alternative would be to grab all possible unprocessed
1538
# file_ids, and then sort by path, and then yield them. That
1539
# might ensure better ordering, in case a caller strictly
1540
# requires parents before children.
1541
for idx, other_extra in enumerate(self._others_extra):
1542
others = sorted(other_extra.itervalues(),
1543
key=lambda x: self._path_to_key(x[0]))
1544
for other_path, other_ie in others:
1545
file_id = other_ie.file_id
1546
# We don't need to check out_of_order_processed here, because
1547
# the lookup_by_file_id will be removing anything processed
1548
# from the extras cache
1549
other_extra.pop(file_id)
1550
other_values = [(None, None) for i in xrange(idx)]
1551
other_values.append((other_path, other_ie))
1552
for alt_idx, alt_extra in enumerate(self._others_extra[idx+1:]):
1553
alt_idx = alt_idx + idx + 1
1554
alt_extra = self._others_extra[alt_idx]
1555
alt_tree = self._other_trees[alt_idx]
1556
other_values.append(self._lookup_by_file_id(
1557
alt_extra, alt_tree, file_id))
1558
yield other_path, file_id, None, other_values
added
removed
bzrlib/tree.py
