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 has_versioned_directories(self):
62
"""Whether this tree can contain explicitly versioned directories.
64
This defaults to True, but some implementations may want to override
69
def changes_from(self, other, want_unchanged=False, specific_files=None,
70
extra_trees=None, require_versioned=False, include_root=False,
71
want_unversioned=False):
72
"""Return a TreeDelta of the changes from other to this tree.
74
:param other: A tree to compare with.
75
:param specific_files: An optional list of file paths to restrict the
76
comparison to. When mapping filenames to ids, all matches in all
77
trees (including optional extra_trees) are used, and all children of
78
matched directories are included.
79
:param want_unchanged: An optional boolean requesting the inclusion of
80
unchanged entries in the result.
81
:param extra_trees: An optional list of additional trees to use when
82
mapping the contents of specific_files (paths) to file_ids.
83
:param require_versioned: An optional boolean (defaults to False). When
84
supplied and True all the 'specific_files' must be versioned, or
85
a PathsNotVersionedError will be thrown.
86
:param want_unversioned: Scan for unversioned paths.
88
The comparison will be performed by an InterTree object looked up on
91
# Martin observes that Tree.changes_from returns a TreeDelta and this
92
# may confuse people, because the class name of the returned object is
93
# a synonym of the object referenced in the method name.
94
return InterTree.get(other, self).compare(
95
want_unchanged=want_unchanged,
96
specific_files=specific_files,
97
extra_trees=extra_trees,
98
require_versioned=require_versioned,
99
include_root=include_root,
100
want_unversioned=want_unversioned,
103
def iter_changes(self, from_tree, include_unchanged=False,
104
specific_files=None, pb=None, extra_trees=None,
105
require_versioned=True, want_unversioned=False):
106
"""See InterTree.iter_changes"""
107
intertree = InterTree.get(from_tree, self)
108
return intertree.iter_changes(include_unchanged, specific_files, pb,
109
extra_trees, require_versioned, want_unversioned=want_unversioned)
112
"""Get a list of the conflicts in the tree.
114
Each conflict is an instance of bzrlib.conflicts.Conflict.
116
return _mod_conflicts.ConflictList()
119
"""For trees that can have unversioned files, return all such paths."""
122
def get_parent_ids(self):
123
"""Get the parent ids for this tree.
125
:return: a list of parent ids. [] is returned to indicate
126
a tree with no parents.
127
:raises: BzrError if the parents are not known.
129
raise NotImplementedError(self.get_parent_ids)
131
def has_filename(self, filename):
132
"""True if the tree has given filename."""
133
raise NotImplementedError(self.has_filename)
135
def has_id(self, file_id):
136
raise NotImplementedError(self.has_id)
138
@deprecated_method(deprecated_in((2, 4, 0)))
139
def __contains__(self, file_id):
140
return self.has_id(file_id)
142
def has_or_had_id(self, file_id):
143
raise NotImplementedError(self.has_or_had_id)
145
def is_ignored(self, filename):
146
"""Check whether the filename is ignored by this tree.
148
:param filename: The relative filename within the tree.
149
:return: True if the filename is ignored.
153
def all_file_ids(self):
154
"""Iterate through all file ids, including ids for missing files."""
155
raise NotImplementedError(self.all_file_ids)
157
def id2path(self, file_id):
158
"""Return the path for a file id.
162
raise NotImplementedError(self.id2path)
164
def iter_entries_by_dir(self, specific_file_ids=None, yield_parents=False):
165
"""Walk the tree in 'by_dir' order.
167
This will yield each entry in the tree as a (path, entry) tuple.
168
The order that they are yielded is:
170
Directories are walked in a depth-first lexicographical order,
171
however, whenever a directory is reached, all of its direct child
172
nodes are yielded in lexicographical order before yielding the
175
For example, in the tree::
185
The yield order (ignoring root) would be::
187
a, f, a/b, a/d, a/b/c, a/d/e, f/g
189
:param yield_parents: If True, yield the parents from the root leading
190
down to specific_file_ids that have been requested. This has no
191
impact if specific_file_ids is None.
193
raise NotImplementedError(self.iter_entries_by_dir)
195
def list_files(self, include_root=False, from_dir=None, recursive=True):
196
"""List all files in this tree.
198
:param include_root: Whether to include the entry for the tree root
199
:param from_dir: Directory under which to list files
200
:param recursive: Whether to list files recursively
201
:return: iterator over tuples of (path, versioned, kind, file_id,
204
raise NotImplementedError(self.list_files)
206
def iter_references(self):
207
if self.supports_tree_reference():
208
for path, entry in self.iter_entries_by_dir():
209
if entry.kind == 'tree-reference':
210
yield path, entry.file_id
212
def kind(self, file_id):
213
raise NotImplementedError("Tree subclass %s must implement kind"
214
% self.__class__.__name__)
216
def stored_kind(self, file_id):
217
"""File kind stored for this file_id.
219
May not match kind on disk for working trees. Always available
220
for versioned files, even when the file itself is missing.
222
return self.kind(file_id)
224
def path_content_summary(self, path):
225
"""Get a summary of the information about path.
227
All the attributes returned are for the canonical form, not the
228
convenient form (if content filters are in use.)
230
:param path: A relative path within the tree.
231
:return: A tuple containing kind, size, exec, sha1-or-link.
232
Kind is always present (see tree.kind()).
233
size is present if kind is file and the size of the
234
canonical form can be cheaply determined, None otherwise.
235
exec is None unless kind is file and the platform supports the 'x'
237
sha1-or-link is the link target if kind is symlink, or the sha1 if
238
it can be obtained without reading the file.
240
raise NotImplementedError(self.path_content_summary)
242
def get_reference_revision(self, file_id, path=None):
243
raise NotImplementedError("Tree subclass %s must implement "
244
"get_reference_revision"
245
% self.__class__.__name__)
247
def _comparison_data(self, entry, path):
248
"""Return a tuple of kind, executable, stat_value for a file.
250
entry may be None if there is no inventory entry for the file, but
251
path must always be supplied.
253
kind is None if there is no file present (even if an inventory id is
254
present). executable is False for non-file entries.
256
raise NotImplementedError(self._comparison_data)
258
def _file_size(self, entry, stat_value):
259
raise NotImplementedError(self._file_size)
261
def get_file(self, file_id, path=None):
262
"""Return a file object for the file file_id in the tree.
264
If both file_id and path are defined, it is implementation defined as
265
to which one is used.
267
raise NotImplementedError(self.get_file)
269
def get_file_with_stat(self, file_id, path=None):
270
"""Get a file handle and stat object for file_id.
272
The default implementation returns (self.get_file, None) for backwards
275
:param file_id: The file id to read.
276
:param path: The path of the file, if it is known.
277
:return: A tuple (file_handle, stat_value_or_None). If the tree has
278
no stat facility, or need for a stat cache feedback during commit,
279
it may return None for the second element of the tuple.
281
return (self.get_file(file_id, path), None)
283
def get_file_text(self, file_id, path=None):
284
"""Return the byte content of a file.
286
:param file_id: The file_id of the file.
287
:param path: The path of the file.
289
If both file_id and path are supplied, an implementation may use
292
:returns: A single byte string for the whole file.
294
my_file = self.get_file(file_id, path)
296
return my_file.read()
300
def get_file_lines(self, file_id, path=None):
301
"""Return the content of a file, as lines.
303
:param file_id: The file_id of the file.
304
:param path: The path of the file.
306
If both file_id and path are supplied, an implementation may use
309
return osutils.split_lines(self.get_file_text(file_id, path))
311
def get_file_verifier(self, file_id, path=None, stat_value=None):
312
"""Return a verifier for a file.
314
The default implementation returns a sha1.
316
:param file_id: The handle for this file.
317
:param path: The path that this file can be found at.
318
These must point to the same object.
319
:param stat_value: Optional stat value for the object
320
:return: Tuple with verifier name and verifier data
322
return ("SHA1", self.get_file_sha1(file_id, path=path,
323
stat_value=stat_value))
325
def get_file_sha1(self, file_id, path=None, stat_value=None):
326
"""Return the SHA1 file for a file.
328
:note: callers should use get_file_verifier instead
329
where possible, as the underlying repository implementation may
330
have quicker access to a non-sha1 verifier.
332
:param file_id: The handle for this file.
333
:param path: The path that this file can be found at.
334
These must point to the same object.
335
:param stat_value: Optional stat value for the object
337
raise NotImplementedError(self.get_file_sha1)
339
def get_file_mtime(self, file_id, path=None):
340
"""Return the modification time for a file.
342
:param file_id: The handle for this file.
343
:param path: The path that this file can be found at.
344
These must point to the same object.
346
raise NotImplementedError(self.get_file_mtime)
348
def get_file_size(self, file_id):
349
"""Return the size of a file in bytes.
351
This applies only to regular files. If invoked on directories or
352
symlinks, it will return None.
353
:param file_id: The file-id of the file
355
raise NotImplementedError(self.get_file_size)
357
def get_file_by_path(self, path):
358
raise NotImplementedError(self.get_file_by_path)
360
def is_executable(self, file_id, path=None):
361
"""Check if a file is executable.
363
:param file_id: The handle for this file.
364
:param path: The path that this file can be found at.
365
These must point to the same object.
367
raise NotImplementedError(self.is_executable)
369
def iter_files_bytes(self, desired_files):
370
"""Iterate through file contents.
372
Files will not necessarily be returned in the order they occur in
373
desired_files. No specific order is guaranteed.
375
Yields pairs of identifier, bytes_iterator. identifier is an opaque
376
value supplied by the caller as part of desired_files. It should
377
uniquely identify the file version in the caller's context. (Examples:
378
an index number or a TreeTransform trans_id.)
380
bytes_iterator is an iterable of bytestrings for the file. The
381
kind of iterable and length of the bytestrings are unspecified, but for
382
this implementation, it is a tuple containing a single bytestring with
383
the complete text of the file.
385
:param desired_files: a list of (file_id, identifier) pairs
387
for file_id, identifier in desired_files:
388
# We wrap the string in a tuple so that we can return an iterable
389
# of bytestrings. (Technically, a bytestring is also an iterable
390
# of bytestrings, but iterating through each character is not
392
cur_file = (self.get_file_text(file_id),)
393
yield identifier, cur_file
395
def get_symlink_target(self, file_id, path=None):
396
"""Get the target for a given file_id.
398
It is assumed that the caller already knows that file_id is referencing
400
:param file_id: Handle for the symlink entry.
401
:param path: The path of the file.
402
If both file_id and path are supplied, an implementation may use
404
:return: The path the symlink points to.
406
raise NotImplementedError(self.get_symlink_target)
408
def get_root_id(self):
409
"""Return the file_id for the root of this tree."""
410
raise NotImplementedError(self.get_root_id)
412
def annotate_iter(self, file_id,
413
default_revision=_mod_revision.CURRENT_REVISION):
414
"""Return an iterator of revision_id, line tuples.
416
For working trees (and mutable trees in general), the special
417
revision_id 'current:' will be used for lines that are new in this
418
tree, e.g. uncommitted changes.
419
:param file_id: The file to produce an annotated version from
420
:param default_revision: For lines that don't match a basis, mark them
421
with this revision id. Not all implementations will make use of
424
raise NotImplementedError(self.annotate_iter)
426
def _get_plan_merge_data(self, file_id, other, base):
427
from bzrlib import versionedfile
428
vf = versionedfile._PlanMergeVersionedFile(file_id)
429
last_revision_a = self._get_file_revision(file_id, vf, 'this:')
430
last_revision_b = other._get_file_revision(file_id, vf, 'other:')
432
last_revision_base = None
434
last_revision_base = base._get_file_revision(file_id, vf, 'base:')
435
return vf, last_revision_a, last_revision_b, last_revision_base
437
def plan_file_merge(self, file_id, other, base=None):
438
"""Generate a merge plan based on annotations.
440
If the file contains uncommitted changes in this tree, they will be
441
attributed to the 'current:' pseudo-revision. If the file contains
442
uncommitted changes in the other tree, they will be assigned to the
443
'other:' pseudo-revision.
445
data = self._get_plan_merge_data(file_id, other, base)
446
vf, last_revision_a, last_revision_b, last_revision_base = data
447
return vf.plan_merge(last_revision_a, last_revision_b,
450
def plan_file_lca_merge(self, file_id, other, base=None):
451
"""Generate a merge plan based lca-newness.
453
If the file contains uncommitted changes in this tree, they will be
454
attributed to the 'current:' pseudo-revision. If the file contains
455
uncommitted changes in the other tree, they will be assigned to the
456
'other:' pseudo-revision.
458
data = self._get_plan_merge_data(file_id, other, base)
459
vf, last_revision_a, last_revision_b, last_revision_base = data
460
return vf.plan_lca_merge(last_revision_a, last_revision_b,
463
def _iter_parent_trees(self):
464
"""Iterate through parent trees, defaulting to Tree.revision_tree."""
465
for revision_id in self.get_parent_ids():
467
yield self.revision_tree(revision_id)
468
except errors.NoSuchRevisionInTree:
469
yield self.repository.revision_tree(revision_id)
471
def _get_file_revision(self, file_id, vf, tree_revision):
472
"""Ensure that file_id, tree_revision is in vf to plan the merge."""
474
if getattr(self, '_repository', None) is None:
475
last_revision = tree_revision
476
parent_keys = [(file_id, t.get_file_revision(file_id)) for t in
477
self._iter_parent_trees()]
478
vf.add_lines((file_id, last_revision), parent_keys,
479
self.get_file_lines(file_id))
480
repo = self.branch.repository
483
last_revision = self.get_file_revision(file_id)
484
base_vf = self._repository.texts
485
if base_vf not in vf.fallback_versionedfiles:
486
vf.fallback_versionedfiles.append(base_vf)
489
def _check_retrieved(self, ie, f):
492
fp = osutils.fingerprint_file(f)
495
if ie.text_size is not None:
496
if ie.text_size != fp['size']:
497
raise errors.BzrError(
498
"mismatched size for file %r in %r" %
499
(ie.file_id, self._store),
500
["inventory expects %d bytes" % ie.text_size,
501
"file is actually %d bytes" % fp['size'],
502
"store is probably damaged/corrupt"])
504
if ie.text_sha1 != fp['sha1']:
505
raise errors.BzrError("wrong SHA-1 for file %r in %r" %
506
(ie.file_id, self._store),
507
["inventory expects %s" % ie.text_sha1,
508
"file is actually %s" % fp['sha1'],
509
"store is probably damaged/corrupt"])
511
def path2id(self, path):
512
"""Return the id for path in this tree."""
513
raise NotImplementedError(self.path2id)
515
def paths2ids(self, paths, trees=[], require_versioned=True):
516
"""Return all the ids that can be reached by walking from paths.
518
Each path is looked up in this tree and any extras provided in
519
trees, and this is repeated recursively: the children in an extra tree
520
of a directory that has been renamed under a provided path in this tree
521
are all returned, even if none exist under a provided path in this
522
tree, and vice versa.
524
:param paths: An iterable of paths to start converting to ids from.
525
Alternatively, if paths is None, no ids should be calculated and None
526
will be returned. This is offered to make calling the api unconditional
527
for code that *might* take a list of files.
528
:param trees: Additional trees to consider.
529
:param require_versioned: If False, do not raise NotVersionedError if
530
an element of paths is not versioned in this tree and all of trees.
532
return find_ids_across_trees(paths, [self] + list(trees), require_versioned)
534
def iter_children(self, file_id):
535
entry = self.iter_entries_by_dir([file_id]).next()[1]
536
for child in getattr(entry, 'children', {}).itervalues():
540
"""Lock this tree for multiple read only operations.
542
:return: A bzrlib.lock.LogicalLockResult.
546
def revision_tree(self, revision_id):
547
"""Obtain a revision tree for the revision revision_id.
549
The intention of this method is to allow access to possibly cached
550
tree data. Implementors of this method should raise NoSuchRevision if
551
the tree is not locally available, even if they could obtain the
552
tree via a repository or some other means. Callers are responsible
553
for finding the ultimate source for a revision tree.
555
:param revision_id: The revision_id of the requested tree.
557
:raises: NoSuchRevision if the tree cannot be obtained.
559
raise errors.NoSuchRevisionInTree(self, revision_id)
562
"""What files are present in this tree and unknown.
564
:return: an iterator over the unknown files.
571
def filter_unversioned_files(self, paths):
572
"""Filter out paths that are versioned.
574
:return: set of paths.
576
raise NotImplementedError(self.filter_unversioned_files)
578
def walkdirs(self, prefix=""):
579
"""Walk the contents of this tree from path down.
581
This yields all the data about the contents of a directory at a time.
582
After each directory has been yielded, if the caller has mutated the
583
list to exclude some directories, they are then not descended into.
585
The data yielded is of the form:
586
((directory-relpath, directory-path-from-root, directory-fileid),
587
[(relpath, basename, kind, lstat, path_from_tree_root, file_id,
588
versioned_kind), ...]),
589
- directory-relpath is the containing dirs relpath from prefix
590
- directory-path-from-root is the containing dirs path from /
591
- directory-fileid is the id of the directory if it is versioned.
592
- relpath is the relative path within the subtree being walked.
593
- basename is the basename
594
- kind is the kind of the file now. If unknonwn then the file is not
595
present within the tree - but it may be recorded as versioned. See
597
- lstat is the stat data *if* the file was statted.
598
- path_from_tree_root is the path from the root of the tree.
599
- file_id is the file_id if the entry is versioned.
600
- versioned_kind is the kind of the file as last recorded in the
601
versioning system. If 'unknown' the file is not versioned.
602
One of 'kind' and 'versioned_kind' must not be 'unknown'.
604
:param prefix: Start walking from prefix within the tree rather than
605
at the root. This allows one to walk a subtree but get paths that are
606
relative to a tree rooted higher up.
607
:return: an iterator over the directory data.
609
raise NotImplementedError(self.walkdirs)
611
def supports_content_filtering(self):
614
def _content_filter_stack(self, path=None, file_id=None):
615
"""The stack of content filters for a path if filtering is supported.
617
Readers will be applied in first-to-last order.
618
Writers will be applied in last-to-first order.
619
Either the path or the file-id needs to be provided.
621
:param path: path relative to the root of the tree
623
:param file_id: file_id or None if unknown
624
:return: the list of filters - [] if there are none
626
filter_pref_names = filters._get_registered_names()
627
if len(filter_pref_names) == 0:
630
path = self.id2path(file_id)
631
prefs = self.iter_search_rules([path], filter_pref_names).next()
632
stk = filters._get_filter_stack_for(prefs)
633
if 'filters' in debug.debug_flags:
634
trace.note("*** %s content-filter: %s => %r" % (path,prefs,stk))
637
def _content_filter_stack_provider(self):
638
"""A function that returns a stack of ContentFilters.
640
The function takes a path (relative to the top of the tree) and a
641
file-id as parameters.
643
:return: None if content filtering is not supported by this tree.
645
if self.supports_content_filtering():
646
return lambda path, file_id: \
647
self._content_filter_stack(path, file_id)
651
def iter_search_rules(self, path_names, pref_names=None,
652
_default_searcher=None):
653
"""Find the preferences for filenames in a tree.
655
:param path_names: an iterable of paths to find attributes for.
656
Paths are given relative to the root of the tree.
657
:param pref_names: the list of preferences to lookup - None for all
658
:param _default_searcher: private parameter to assist testing - don't use
659
:return: an iterator of tuple sequences, one per path-name.
660
See _RulesSearcher.get_items for details on the tuple sequence.
662
if _default_searcher is None:
663
_default_searcher = rules._per_user_searcher
664
searcher = self._get_rules_searcher(_default_searcher)
665
if searcher is not None:
666
if pref_names is not None:
667
for path in path_names:
668
yield searcher.get_selected_items(path, pref_names)
670
for path in path_names:
671
yield searcher.get_items(path)
673
def _get_rules_searcher(self, default_searcher):
674
"""Get the RulesSearcher for this tree given the default one."""
675
searcher = default_searcher
679
class InventoryTree(Tree):
680
"""A tree that relies on an inventory for its metadata.
682
Trees contain an `Inventory` object, and also know how to retrieve
683
file texts mentioned in the inventory, either from a working
684
directory or from a store.
686
It is possible for trees to contain files that are not described
687
in their inventory or vice versa; for this use `filenames()`.
689
Subclasses should set the _inventory attribute, which is considered
690
private to external API users.
693
def get_canonical_inventory_paths(self, paths):
694
"""Like get_canonical_inventory_path() but works on multiple items.
696
:param paths: A sequence of paths relative to the root of the tree.
697
:return: A list of paths, with each item the corresponding input path
698
adjusted to account for existing elements that match case
701
return list(self._yield_canonical_inventory_paths(paths))
703
def get_canonical_inventory_path(self, path):
704
"""Returns the first inventory item that case-insensitively matches path.
706
If a path matches exactly, it is returned. If no path matches exactly
707
but more than one path matches case-insensitively, it is implementation
708
defined which is returned.
710
If no path matches case-insensitively, the input path is returned, but
711
with as many path entries that do exist changed to their canonical
714
If you need to resolve many names from the same tree, you should
715
use get_canonical_inventory_paths() to avoid O(N) behaviour.
717
:param path: A paths relative to the root of the tree.
718
:return: The input path adjusted to account for existing elements
719
that match case insensitively.
721
return self._yield_canonical_inventory_paths([path]).next()
723
def _yield_canonical_inventory_paths(self, paths):
725
# First, if the path as specified exists exactly, just use it.
726
if self.path2id(path) is not None:
730
cur_id = self.get_root_id()
732
bit_iter = iter(path.split("/"))
736
for child in self.iter_children(cur_id):
738
# XXX: it seem like if the child is known to be in the
739
# tree, we shouldn't need to go from its id back to
740
# its path -- mbp 2010-02-11
742
# XXX: it seems like we could be more efficient
743
# by just directly looking up the original name and
744
# only then searching all children; also by not
745
# chopping paths so much. -- mbp 2010-02-11
746
child_base = os.path.basename(self.id2path(child))
747
if (child_base == elt):
748
# if we found an exact match, we can stop now; if
749
# we found an approximate match we need to keep
750
# searching because there might be an exact match
753
new_path = osutils.pathjoin(cur_path, child_base)
755
elif child_base.lower() == lelt:
757
new_path = osutils.pathjoin(cur_path, child_base)
758
except errors.NoSuchId:
759
# before a change is committed we can see this error...
764
# got to the end of this directory and no entries matched.
765
# Return what matched so far, plus the rest as specified.
766
cur_path = osutils.pathjoin(cur_path, elt, *list(bit_iter))
771
def _get_inventory(self):
772
return self._inventory
774
inventory = property(_get_inventory,
775
doc="Inventory of this Tree")
778
def path2id(self, path):
779
"""Return the id for path in this tree."""
780
return self._inventory.path2id(path)
782
def id2path(self, file_id):
783
"""Return the path for a file id.
787
return self.inventory.id2path(file_id)
789
def has_id(self, file_id):
790
return self.inventory.has_id(file_id)
792
def has_or_had_id(self, file_id):
793
return self.inventory.has_id(file_id)
795
def all_file_ids(self):
796
return set(self.inventory)
798
@deprecated_method(deprecated_in((2, 4, 0)))
800
return iter(self.inventory)
802
def filter_unversioned_files(self, paths):
803
"""Filter out paths that are versioned.
805
:return: set of paths.
807
# NB: we specifically *don't* call self.has_filename, because for
808
# WorkingTrees that can indicate files that exist on disk but that
810
pred = self.inventory.has_filename
811
return set((p for p in paths if not pred(p)))
814
def iter_entries_by_dir(self, specific_file_ids=None, yield_parents=False):
815
"""Walk the tree in 'by_dir' order.
817
This will yield each entry in the tree as a (path, entry) tuple.
818
The order that they are yielded is:
820
See Tree.iter_entries_by_dir for details.
822
:param yield_parents: If True, yield the parents from the root leading
823
down to specific_file_ids that have been requested. This has no
824
impact if specific_file_ids is None.
826
return self.inventory.iter_entries_by_dir(
827
specific_file_ids=specific_file_ids, yield_parents=yield_parents)
829
def get_file_by_path(self, path):
830
return self.get_file(self._inventory.path2id(path), path)
833
def find_ids_across_trees(filenames, trees, require_versioned=True):
834
"""Find the ids corresponding to specified filenames.
836
All matches in all trees will be used, and all children of matched
837
directories will be used.
839
:param filenames: The filenames to find file_ids for (if None, returns
841
:param trees: The trees to find file_ids within
842
:param require_versioned: if true, all specified filenames must occur in
844
:return: a set of file ids for the specified filenames and their children.
848
specified_path_ids = _find_ids_across_trees(filenames, trees,
850
return _find_children_across_trees(specified_path_ids, trees)
853
def _find_ids_across_trees(filenames, trees, require_versioned):
854
"""Find the ids corresponding to specified filenames.
856
All matches in all trees will be used, but subdirectories are not scanned.
858
:param filenames: The filenames to find file_ids for
859
:param trees: The trees to find file_ids within
860
:param require_versioned: if true, all specified filenames must occur in
862
:return: a set of file ids for the specified filenames
865
interesting_ids = set()
866
for tree_path in filenames:
869
file_id = tree.path2id(tree_path)
870
if file_id is not None:
871
interesting_ids.add(file_id)
874
not_versioned.append(tree_path)
875
if len(not_versioned) > 0 and require_versioned:
876
raise errors.PathsNotVersionedError(not_versioned)
877
return interesting_ids
880
def _find_children_across_trees(specified_ids, trees):
881
"""Return a set including specified ids and their children.
883
All matches in all trees will be used.
885
:param trees: The trees to find file_ids within
886
:return: a set containing all specified ids and their children
888
interesting_ids = set(specified_ids)
889
pending = interesting_ids
890
# now handle children of interesting ids
891
# we loop so that we handle all children of each id in both trees
892
while len(pending) > 0:
894
for file_id in pending:
896
if not tree.has_or_had_id(file_id):
898
for child_id in tree.iter_children(file_id):
899
if child_id not in interesting_ids:
900
new_pending.add(child_id)
901
interesting_ids.update(new_pending)
902
pending = new_pending
903
return interesting_ids
906
class InterTree(InterObject):
907
"""This class represents operations taking place between two Trees.
909
Its instances have methods like 'compare' and contain references to the
910
source and target trees these operations are to be carried out on.
912
Clients of bzrlib should not need to use InterTree directly, rather they
913
should use the convenience methods on Tree such as 'Tree.compare()' which
914
will pass through to InterTree as appropriate.
917
# Formats that will be used to test this InterTree. If both are
918
# None, this InterTree will not be tested (e.g. because a complex
920
_matching_from_tree_format = None
921
_matching_to_tree_format = None
926
def is_compatible(kls, source, target):
927
# The default implementation is naive and uses the public API, so
928
# it works for all trees.
931
def _changes_from_entries(self, source_entry, target_entry,
932
source_path=None, target_path=None):
933
"""Generate a iter_changes tuple between source_entry and target_entry.
935
:param source_entry: An inventory entry from self.source, or None.
936
:param target_entry: An inventory entry from self.target, or None.
937
:param source_path: The path of source_entry, if known. If not known
938
it will be looked up.
939
:param target_path: The path of target_entry, if known. If not known
940
it will be looked up.
941
:return: A tuple, item 0 of which is an iter_changes result tuple, and
942
item 1 is True if there are any changes in the result tuple.
944
if source_entry is None:
945
if target_entry is None:
947
file_id = target_entry.file_id
949
file_id = source_entry.file_id
950
if source_entry is not None:
951
source_versioned = True
952
source_name = source_entry.name
953
source_parent = source_entry.parent_id
954
if source_path is None:
955
source_path = self.source.id2path(file_id)
956
source_kind, source_executable, source_stat = \
957
self.source._comparison_data(source_entry, source_path)
959
source_versioned = False
963
source_executable = None
964
if target_entry is not None:
965
target_versioned = True
966
target_name = target_entry.name
967
target_parent = target_entry.parent_id
968
if target_path is None:
969
target_path = self.target.id2path(file_id)
970
target_kind, target_executable, target_stat = \
971
self.target._comparison_data(target_entry, target_path)
973
target_versioned = False
977
target_executable = None
978
versioned = (source_versioned, target_versioned)
979
kind = (source_kind, target_kind)
980
changed_content = False
981
if source_kind != target_kind:
982
changed_content = True
983
elif source_kind == 'file':
984
if not self.file_content_matches(file_id, file_id, source_path,
985
target_path, source_stat, target_stat):
986
changed_content = True
987
elif source_kind == 'symlink':
988
if (self.source.get_symlink_target(file_id) !=
989
self.target.get_symlink_target(file_id)):
990
changed_content = True
991
# XXX: Yes, the indentation below is wrong. But fixing it broke
992
# test_merge.TestMergerEntriesLCAOnDisk.
993
# test_nested_tree_subtree_renamed_and_modified. We'll wait for
994
# the fix from bzr.dev -- vila 2009026
995
elif source_kind == 'tree-reference':
996
if (self.source.get_reference_revision(file_id, source_path)
997
!= self.target.get_reference_revision(file_id, target_path)):
998
changed_content = True
999
parent = (source_parent, target_parent)
1000
name = (source_name, target_name)
1001
executable = (source_executable, target_executable)
1002
if (changed_content is not False or versioned[0] != versioned[1]
1003
or parent[0] != parent[1] or name[0] != name[1] or
1004
executable[0] != executable[1]):
1008
return (file_id, (source_path, target_path), changed_content,
1009
versioned, parent, name, kind, executable), changes
1012
def compare(self, want_unchanged=False, specific_files=None,
1013
extra_trees=None, require_versioned=False, include_root=False,
1014
want_unversioned=False):
1015
"""Return the changes from source to target.
1017
:return: A TreeDelta.
1018
:param specific_files: An optional list of file paths to restrict the
1019
comparison to. When mapping filenames to ids, all matches in all
1020
trees (including optional extra_trees) are used, and all children of
1021
matched directories are included.
1022
:param want_unchanged: An optional boolean requesting the inclusion of
1023
unchanged entries in the result.
1024
:param extra_trees: An optional list of additional trees to use when
1025
mapping the contents of specific_files (paths) to file_ids.
1026
:param require_versioned: An optional boolean (defaults to False). When
1027
supplied and True all the 'specific_files' must be versioned, or
1028
a PathsNotVersionedError will be thrown.
1029
:param want_unversioned: Scan for unversioned paths.
1031
trees = (self.source,)
1032
if extra_trees is not None:
1033
trees = trees + tuple(extra_trees)
1034
# target is usually the newer tree:
1035
specific_file_ids = self.target.paths2ids(specific_files, trees,
1036
require_versioned=require_versioned)
1037
if specific_files and not specific_file_ids:
1038
# All files are unversioned, so just return an empty delta
1039
# _compare_trees would think we want a complete delta
1040
result = delta.TreeDelta()
1041
fake_entry = inventory.InventoryFile('unused', 'unused', 'unused')
1042
result.unversioned = [(path, None,
1043
self.target._comparison_data(fake_entry, path)[0]) for path in
1046
return delta._compare_trees(self.source, self.target, want_unchanged,
1047
specific_files, include_root, extra_trees=extra_trees,
1048
require_versioned=require_versioned,
1049
want_unversioned=want_unversioned)
1051
def iter_changes(self, include_unchanged=False,
1052
specific_files=None, pb=None, extra_trees=[],
1053
require_versioned=True, want_unversioned=False):
1054
"""Generate an iterator of changes between trees.
1056
A tuple is returned:
1057
(file_id, (path_in_source, path_in_target),
1058
changed_content, versioned, parent, name, kind,
1061
Changed_content is True if the file's content has changed. This
1062
includes changes to its kind, and to a symlink's target.
1064
versioned, parent, name, kind, executable are tuples of (from, to).
1065
If a file is missing in a tree, its kind is None.
1067
Iteration is done in parent-to-child order, relative to the target
1070
There is no guarantee that all paths are in sorted order: the
1071
requirement to expand the search due to renames may result in children
1072
that should be found early being found late in the search, after
1073
lexically later results have been returned.
1074
:param require_versioned: Raise errors.PathsNotVersionedError if a
1075
path in the specific_files list is not versioned in one of
1076
source, target or extra_trees.
1077
:param specific_files: An optional list of file paths to restrict the
1078
comparison to. When mapping filenames to ids, all matches in all
1079
trees (including optional extra_trees) are used, and all children
1080
of matched directories are included. The parents in the target tree
1081
of the specific files up to and including the root of the tree are
1082
always evaluated for changes too.
1083
:param want_unversioned: Should unversioned files be returned in the
1084
output. An unversioned file is defined as one with (False, False)
1085
for the versioned pair.
1087
lookup_trees = [self.source]
1089
lookup_trees.extend(extra_trees)
1090
# The ids of items we need to examine to insure delta consistency.
1091
precise_file_ids = set()
1092
changed_file_ids = []
1093
if specific_files == []:
1094
specific_file_ids = []
1096
specific_file_ids = self.target.paths2ids(specific_files,
1097
lookup_trees, require_versioned=require_versioned)
1098
if specific_files is not None:
1099
# reparented or added entries must have their parents included
1100
# so that valid deltas can be created. The seen_parents set
1101
# tracks the parents that we need to have.
1102
# The seen_dirs set tracks directory entries we've yielded.
1103
# After outputting version object in to_entries we set difference
1104
# the two seen sets and start checking parents.
1105
seen_parents = set()
1107
if want_unversioned:
1108
all_unversioned = sorted([(p.split('/'), p) for p in
1109
self.target.extras()
1110
if specific_files is None or
1111
osutils.is_inside_any(specific_files, p)])
1112
all_unversioned = collections.deque(all_unversioned)
1114
all_unversioned = collections.deque()
1116
from_entries_by_dir = list(self.source.iter_entries_by_dir(
1117
specific_file_ids=specific_file_ids))
1118
from_data = dict((e.file_id, (p, e)) for p, e in from_entries_by_dir)
1119
to_entries_by_dir = list(self.target.iter_entries_by_dir(
1120
specific_file_ids=specific_file_ids))
1121
num_entries = len(from_entries_by_dir) + len(to_entries_by_dir)
1123
# the unversioned path lookup only occurs on real trees - where there
1124
# can be extras. So the fake_entry is solely used to look up
1125
# executable it values when execute is not supported.
1126
fake_entry = inventory.InventoryFile('unused', 'unused', 'unused')
1127
for target_path, target_entry in to_entries_by_dir:
1128
while (all_unversioned and
1129
all_unversioned[0][0] < target_path.split('/')):
1130
unversioned_path = all_unversioned.popleft()
1131
target_kind, target_executable, target_stat = \
1132
self.target._comparison_data(fake_entry, unversioned_path[1])
1133
yield (None, (None, unversioned_path[1]), True, (False, False),
1135
(None, unversioned_path[0][-1]),
1136
(None, target_kind),
1137
(None, target_executable))
1138
source_path, source_entry = from_data.get(target_entry.file_id,
1140
result, changes = self._changes_from_entries(source_entry,
1141
target_entry, source_path=source_path, target_path=target_path)
1142
to_paths[result[0]] = result[1][1]
1147
pb.update('comparing files', entry_count, num_entries)
1148
if changes or include_unchanged:
1149
if specific_file_ids is not None:
1150
new_parent_id = result[4][1]
1151
precise_file_ids.add(new_parent_id)
1152
changed_file_ids.append(result[0])
1154
# Ensure correct behaviour for reparented/added specific files.
1155
if specific_files is not None:
1156
# Record output dirs
1157
if result[6][1] == 'directory':
1158
seen_dirs.add(result[0])
1159
# Record parents of reparented/added entries.
1160
versioned = result[3]
1162
if not versioned[0] or parents[0] != parents[1]:
1163
seen_parents.add(parents[1])
1164
while all_unversioned:
1165
# yield any trailing unversioned paths
1166
unversioned_path = all_unversioned.popleft()
1167
to_kind, to_executable, to_stat = \
1168
self.target._comparison_data(fake_entry, unversioned_path[1])
1169
yield (None, (None, unversioned_path[1]), True, (False, False),
1171
(None, unversioned_path[0][-1]),
1173
(None, to_executable))
1174
# Yield all remaining source paths
1175
for path, from_entry in from_entries_by_dir:
1176
file_id = from_entry.file_id
1177
if file_id in to_paths:
1180
if not self.target.has_id(file_id):
1181
# common case - paths we have not emitted are not present in
1185
to_path = self.target.id2path(file_id)
1188
pb.update('comparing files', entry_count, num_entries)
1189
versioned = (True, False)
1190
parent = (from_entry.parent_id, None)
1191
name = (from_entry.name, None)
1192
from_kind, from_executable, stat_value = \
1193
self.source._comparison_data(from_entry, path)
1194
kind = (from_kind, None)
1195
executable = (from_executable, None)
1196
changed_content = from_kind is not None
1197
# the parent's path is necessarily known at this point.
1198
changed_file_ids.append(file_id)
1199
yield(file_id, (path, to_path), changed_content, versioned, parent,
1200
name, kind, executable)
1201
changed_file_ids = set(changed_file_ids)
1202
if specific_file_ids is not None:
1203
for result in self._handle_precise_ids(precise_file_ids,
1207
def _get_entry(self, tree, file_id):
1208
"""Get an inventory entry from a tree, with missing entries as None.
1210
If the tree raises NotImplementedError on accessing .inventory, then
1211
this is worked around using iter_entries_by_dir on just the file id
1214
:param tree: The tree to lookup the entry in.
1215
:param file_id: The file_id to lookup.
1218
inventory = tree.inventory
1219
except NotImplementedError:
1220
# No inventory available.
1222
iterator = tree.iter_entries_by_dir(specific_file_ids=[file_id])
1223
return iterator.next()[1]
1224
except StopIteration:
1228
return inventory[file_id]
1229
except errors.NoSuchId:
1232
def _handle_precise_ids(self, precise_file_ids, changed_file_ids,
1233
discarded_changes=None):
1234
"""Fill out a partial iter_changes to be consistent.
1236
:param precise_file_ids: The file ids of parents that were seen during
1238
:param changed_file_ids: The file ids of already emitted items.
1239
:param discarded_changes: An optional dict of precalculated
1240
iter_changes items which the partial iter_changes had not output
1242
:return: A generator of iter_changes items to output.
1244
# process parents of things that had changed under the users
1245
# requested paths to prevent incorrect paths or parent ids which
1246
# aren't in the tree.
1247
while precise_file_ids:
1248
precise_file_ids.discard(None)
1249
# Don't emit file_ids twice
1250
precise_file_ids.difference_update(changed_file_ids)
1251
if not precise_file_ids:
1253
# If the there was something at a given output path in source, we
1254
# have to include the entry from source in the delta, or we would
1255
# be putting this entry into a used path.
1257
for parent_id in precise_file_ids:
1259
paths.append(self.target.id2path(parent_id))
1260
except errors.NoSuchId:
1261
# This id has been dragged in from the source by delta
1262
# expansion and isn't present in target at all: we don't
1263
# need to check for path collisions on it.
1266
old_id = self.source.path2id(path)
1267
precise_file_ids.add(old_id)
1268
precise_file_ids.discard(None)
1269
current_ids = precise_file_ids
1270
precise_file_ids = set()
1271
# We have to emit all of precise_file_ids that have been altered.
1272
# We may have to output the children of some of those ids if any
1273
# directories have stopped being directories.
1274
for file_id in current_ids:
1276
if discarded_changes:
1277
result = discarded_changes.get(file_id)
1282
old_entry = self._get_entry(self.source, file_id)
1283
new_entry = self._get_entry(self.target, file_id)
1284
result, changes = self._changes_from_entries(
1285
old_entry, new_entry)
1288
# Get this parents parent to examine.
1289
new_parent_id = result[4][1]
1290
precise_file_ids.add(new_parent_id)
1292
if (result[6][0] == 'directory' and
1293
result[6][1] != 'directory'):
1294
# This stopped being a directory, the old children have
1296
if old_entry is None:
1297
# Reusing a discarded change.
1298
old_entry = self._get_entry(self.source, file_id)
1299
for child in old_entry.children.values():
1300
precise_file_ids.add(child.file_id)
1301
changed_file_ids.add(result[0])
1305
def file_content_matches(self, source_file_id, target_file_id,
1306
source_path=None, target_path=None, source_stat=None, target_stat=None):
1307
"""Check if two files are the same in the source and target trees.
1309
This only checks that the contents of the files are the same,
1310
it does not touch anything else.
1312
:param source_file_id: File id of the file in the source tree
1313
:param target_file_id: File id of the file in the target tree
1314
:param source_path: Path of the file in the source tree
1315
:param target_path: Path of the file in the target tree
1316
:param source_stat: Optional stat value of the file in the source tree
1317
:param target_stat: Optional stat value of the file in the target tree
1318
:return: Boolean indicating whether the files have the same contents
1320
source_verifier_kind, source_verifier_data = self.source.get_file_verifier(
1321
source_file_id, source_path, source_stat)
1322
target_verifier_kind, target_verifier_data = self.target.get_file_verifier(
1323
target_file_id, target_path, target_stat)
1324
if source_verifier_kind == target_verifier_kind:
1325
return (source_verifier_data == target_verifier_data)
1326
# Fall back to SHA1 for now
1327
if source_verifier_kind != "SHA1":
1328
source_sha1 = self.source.get_file_sha1(source_file_id,
1329
source_path, source_stat)
1331
source_sha1 = source_verifier_data
1332
if target_verifier_kind != "SHA1":
1333
target_sha1 = self.target.get_file_sha1(target_file_id,
1334
target_path, target_stat)
1336
target_sha1 = target_verifier_data
1337
return (source_sha1 == target_sha1)
1339
InterTree.register_optimiser(InterTree)
1342
class MultiWalker(object):
1343
"""Walk multiple trees simultaneously, getting combined results."""
1345
# Note: This could be written to not assume you can do out-of-order
1346
# lookups. Instead any nodes that don't match in all trees could be
1347
# marked as 'deferred', and then returned in the final cleanup loop.
1348
# For now, I think it is "nicer" to return things as close to the
1349
# "master_tree" order as we can.
1351
def __init__(self, master_tree, other_trees):
1352
"""Create a new MultiWalker.
1354
All trees being walked must implement "iter_entries_by_dir()", such
1355
that they yield (path, object) tuples, where that object will have a
1356
'.file_id' member, that can be used to check equality.
1358
:param master_tree: All trees will be 'slaved' to the master_tree such
1359
that nodes in master_tree will be used as 'first-pass' sync points.
1360
Any nodes that aren't in master_tree will be merged in a second
1362
:param other_trees: A list of other trees to walk simultaneously.
1364
self._master_tree = master_tree
1365
self._other_trees = other_trees
1367
# Keep track of any nodes that were properly processed just out of
1368
# order, that way we don't return them at the end, we don't have to
1369
# track *all* processed file_ids, just the out-of-order ones
1370
self._out_of_order_processed = set()
1373
def _step_one(iterator):
1374
"""Step an iter_entries_by_dir iterator.
1376
:return: (has_more, path, ie)
1377
If has_more is False, path and ie will be None.
1380
path, ie = iterator.next()
1381
except StopIteration:
1382
return False, None, None
1384
return True, path, ie
1387
def _cmp_path_by_dirblock(path1, path2):
1388
"""Compare two paths based on what directory they are in.
1390
This generates a sort order, such that all children of a directory are
1391
sorted together, and grandchildren are in the same order as the
1392
children appear. But all grandchildren come after all children.
1394
:param path1: first path
1395
:param path2: the second path
1396
:return: negative number if ``path1`` comes first,
1397
0 if paths are equal
1398
and a positive number if ``path2`` sorts first
1400
# Shortcut this special case
1403
# This is stolen from _dirstate_helpers_py.py, only switching it to
1404
# Unicode objects. Consider using encode_utf8() and then using the
1405
# optimized versions, or maybe writing optimized unicode versions.
1406
if not isinstance(path1, unicode):
1407
raise TypeError("'path1' must be a unicode string, not %s: %r"
1408
% (type(path1), path1))
1409
if not isinstance(path2, unicode):
1410
raise TypeError("'path2' must be a unicode string, not %s: %r"
1411
% (type(path2), path2))
1412
return cmp(MultiWalker._path_to_key(path1),
1413
MultiWalker._path_to_key(path2))
1416
def _path_to_key(path):
1417
dirname, basename = osutils.split(path)
1418
return (dirname.split(u'/'), basename)
1420
def _lookup_by_file_id(self, extra_entries, other_tree, file_id):
1421
"""Lookup an inventory entry by file_id.
1423
This is called when an entry is missing in the normal order.
1424
Generally this is because a file was either renamed, or it was
1425
deleted/added. If the entry was found in the inventory and not in
1426
extra_entries, it will be added to self._out_of_order_processed
1428
:param extra_entries: A dictionary of {file_id: (path, ie)}. This
1429
should be filled with entries that were found before they were
1430
used. If file_id is present, it will be removed from the
1432
:param other_tree: The Tree to search, in case we didn't find the entry
1434
:param file_id: The file_id to look for
1435
:return: (path, ie) if found or (None, None) if not present.
1437
if file_id in extra_entries:
1438
return extra_entries.pop(file_id)
1439
# TODO: Is id2path better as the first call, or is
1440
# inventory[file_id] better as a first check?
1442
cur_path = other_tree.id2path(file_id)
1443
except errors.NoSuchId:
1445
if cur_path is None:
1448
self._out_of_order_processed.add(file_id)
1449
cur_ie = other_tree.inventory[file_id]
1450
return (cur_path, cur_ie)
1453
"""Match up the values in the different trees."""
1454
for result in self._walk_master_tree():
1456
self._finish_others()
1457
for result in self._walk_others():
1460
def _walk_master_tree(self):
1461
"""First pass, walk all trees in lock-step.
1463
When we are done, all nodes in the master_tree will have been
1464
processed. _other_walkers, _other_entries, and _others_extra will be
1465
set on 'self' for future processing.
1467
# This iterator has the most "inlining" done, because it tends to touch
1468
# every file in the tree, while the others only hit nodes that don't
1470
master_iterator = self._master_tree.iter_entries_by_dir()
1472
other_walkers = [other.iter_entries_by_dir()
1473
for other in self._other_trees]
1474
other_entries = [self._step_one(walker) for walker in other_walkers]
1475
# Track extra nodes in the other trees
1476
others_extra = [{} for i in xrange(len(self._other_trees))]
1478
master_has_more = True
1479
step_one = self._step_one
1480
lookup_by_file_id = self._lookup_by_file_id
1481
out_of_order_processed = self._out_of_order_processed
1483
while master_has_more:
1484
(master_has_more, path, master_ie) = step_one(master_iterator)
1485
if not master_has_more:
1488
file_id = master_ie.file_id
1490
other_values_append = other_values.append
1491
next_other_entries = []
1492
next_other_entries_append = next_other_entries.append
1493
for idx, (other_has_more, other_path, other_ie) in enumerate(other_entries):
1494
if not other_has_more:
1495
other_values_append(lookup_by_file_id(
1496
others_extra[idx], self._other_trees[idx], file_id))
1497
next_other_entries_append((False, None, None))
1498
elif file_id == other_ie.file_id:
1499
# This is the critical code path, as most of the entries
1500
# should match between most trees.
1501
other_values_append((other_path, other_ie))
1502
next_other_entries_append(step_one(other_walkers[idx]))
1504
# This walker did not match, step it until it either
1505
# matches, or we know we are past the current walker.
1506
other_walker = other_walkers[idx]
1507
other_extra = others_extra[idx]
1508
while (other_has_more and
1509
self._cmp_path_by_dirblock(other_path, path) < 0):
1510
other_file_id = other_ie.file_id
1511
if other_file_id not in out_of_order_processed:
1512
other_extra[other_file_id] = (other_path, other_ie)
1513
other_has_more, other_path, other_ie = \
1514
step_one(other_walker)
1515
if other_has_more and other_ie.file_id == file_id:
1516
# We ended up walking to this point, match and step
1518
other_values_append((other_path, other_ie))
1519
other_has_more, other_path, other_ie = \
1520
step_one(other_walker)
1522
# This record isn't in the normal order, see if it
1524
other_values_append(lookup_by_file_id(
1525
other_extra, self._other_trees[idx], file_id))
1526
next_other_entries_append((other_has_more, other_path,
1528
other_entries = next_other_entries
1530
# We've matched all the walkers, yield this datapoint
1531
yield path, file_id, master_ie, other_values
1532
self._other_walkers = other_walkers
1533
self._other_entries = other_entries
1534
self._others_extra = others_extra
1536
def _finish_others(self):
1537
"""Finish walking the other iterators, so we get all entries."""
1538
for idx, info in enumerate(self._other_entries):
1539
other_extra = self._others_extra[idx]
1540
(other_has_more, other_path, other_ie) = info
1541
while other_has_more:
1542
other_file_id = other_ie.file_id
1543
if other_file_id not in self._out_of_order_processed:
1544
other_extra[other_file_id] = (other_path, other_ie)
1545
other_has_more, other_path, other_ie = \
1546
self._step_one(self._other_walkers[idx])
1547
del self._other_entries
1549
def _walk_others(self):
1550
"""Finish up by walking all the 'deferred' nodes."""
1551
# TODO: One alternative would be to grab all possible unprocessed
1552
# file_ids, and then sort by path, and then yield them. That
1553
# might ensure better ordering, in case a caller strictly
1554
# requires parents before children.
1555
for idx, other_extra in enumerate(self._others_extra):
1556
others = sorted(other_extra.itervalues(),
1557
key=lambda x: self._path_to_key(x[0]))
1558
for other_path, other_ie in others:
1559
file_id = other_ie.file_id
1560
# We don't need to check out_of_order_processed here, because
1561
# the lookup_by_file_id will be removing anything processed
1562
# from the extras cache
1563
other_extra.pop(file_id)
1564
other_values = [(None, None) for i in xrange(idx)]
1565
other_values.append((other_path, other_ie))
1566
for alt_idx, alt_extra in enumerate(self._others_extra[idx+1:]):
1567
alt_idx = alt_idx + idx + 1
1568
alt_extra = self._others_extra[alt_idx]
1569
alt_tree = self._other_trees[alt_idx]
1570
other_values.append(self._lookup_by_file_id(
1571
alt_extra, alt_tree, file_id))
1572
yield other_path, file_id, None, other_values
added
removed
bzrlib/tree.py
