165
170
return self.bzrdir.is_control_filename(filename)
168
def iter_entries_by_dir(self, specific_file_ids=None):
173
def iter_entries_by_dir(self, specific_file_ids=None, yield_parents=False):
169
174
"""Walk the tree in 'by_dir' order.
171
This will yield each entry in the tree as a (path, entry) tuple. The
172
order that they are yielded is: the contents of a directory are
173
preceeded by the parent of a directory, and all the contents of a
174
directory are grouped together.
176
This will yield each entry in the tree as a (path, entry) tuple.
177
The order that they are yielded is:
179
Directories are walked in a depth-first lexicographical order,
180
however, whenever a directory is reached, all of its direct child
181
nodes are yielded in lexicographical order before yielding the
184
For example, in the tree::
194
The yield order (ignoring root) would be::
195
a, f, a/b, a/d, a/b/c, a/d/e, f/g
197
:param yield_parents: If True, yield the parents from the root leading
198
down to specific_file_ids that have been requested. This has no
199
impact if specific_file_ids is None.
176
201
return self.inventory.iter_entries_by_dir(
177
specific_file_ids=specific_file_ids)
202
specific_file_ids=specific_file_ids, yield_parents=yield_parents)
179
204
def iter_references(self):
180
for path, entry in self.iter_entries_by_dir():
181
if entry.kind == 'tree-reference':
182
yield path, entry.file_id
205
if self.supports_tree_reference():
206
for path, entry in self.iter_entries_by_dir():
207
if entry.kind == 'tree-reference':
208
yield path, entry.file_id
184
210
def kind(self, file_id):
185
211
raise NotImplementedError("Tree subclass %s must implement kind"
186
212
% self.__class__.__name__)
214
def stored_kind(self, file_id):
215
"""File kind stored for this file_id.
217
May not match kind on disk for working trees. Always available
218
for versioned files, even when the file itself is missing.
220
return self.kind(file_id)
188
222
def path_content_summary(self, path):
189
223
"""Get a summary of the information about path.
225
All the attributes returned are for the canonical form, not the
226
convenient form (if content filters are in use.)
191
228
:param path: A relative path within the tree.
192
229
:return: A tuple containing kind, size, exec, sha1-or-link.
193
230
Kind is always present (see tree.kind()).
194
size is present if kind is file, None otherwise.
231
size is present if kind is file and the size of the
232
canonical form can be cheaply determined, None otherwise.
195
233
exec is None unless kind is file and the platform supports the 'x'
197
235
sha1-or-link is the link target if kind is symlink, or the sha1 if
278
363
raise NotImplementedError(self.get_symlink_target)
280
def annotate_iter(self, file_id):
365
def get_canonical_inventory_paths(self, paths):
366
"""Like get_canonical_inventory_path() but works on multiple items.
368
:param paths: A sequence of paths relative to the root of the tree.
369
:return: A list of paths, with each item the corresponding input path
370
adjusted to account for existing elements that match case
373
return list(self._yield_canonical_inventory_paths(paths))
375
def get_canonical_inventory_path(self, path):
376
"""Returns the first inventory item that case-insensitively matches path.
378
If a path matches exactly, it is returned. If no path matches exactly
379
but more than one path matches case-insensitively, it is implementation
380
defined which is returned.
382
If no path matches case-insensitively, the input path is returned, but
383
with as many path entries that do exist changed to their canonical
386
If you need to resolve many names from the same tree, you should
387
use get_canonical_inventory_paths() to avoid O(N) behaviour.
389
:param path: A paths relative to the root of the tree.
390
:return: The input path adjusted to account for existing elements
391
that match case insensitively.
393
return self._yield_canonical_inventory_paths([path]).next()
395
def _yield_canonical_inventory_paths(self, paths):
397
# First, if the path as specified exists exactly, just use it.
398
if self.path2id(path) is not None:
402
cur_id = self.get_root_id()
404
bit_iter = iter(path.split("/"))
407
for child in self.iter_children(cur_id):
409
child_base = os.path.basename(self.id2path(child))
410
if child_base.lower() == lelt:
412
cur_path = osutils.pathjoin(cur_path, child_base)
415
# before a change is committed we can see this error...
418
# got to the end of this directory and no entries matched.
419
# Return what matched so far, plus the rest as specified.
420
cur_path = osutils.pathjoin(cur_path, elt, *list(bit_iter))
425
def get_root_id(self):
426
"""Return the file_id for the root of this tree."""
427
raise NotImplementedError(self.get_root_id)
429
def annotate_iter(self, file_id,
430
default_revision=_mod_revision.CURRENT_REVISION):
281
431
"""Return an iterator of revision_id, line tuples.
283
433
For working trees (and mutable trees in general), the special
284
434
revision_id 'current:' will be used for lines that are new in this
285
435
tree, e.g. uncommitted changes.
286
436
:param file_id: The file to produce an annotated version from
437
:param default_revision: For lines that don't match a basis, mark them
438
with this revision id. Not all implementations will make use of
288
441
raise NotImplementedError(self.annotate_iter)
290
def plan_file_merge(self, file_id, other):
443
def _get_plan_merge_data(self, file_id, other, base):
444
from bzrlib import versionedfile
445
vf = versionedfile._PlanMergeVersionedFile(file_id)
446
last_revision_a = self._get_file_revision(file_id, vf, 'this:')
447
last_revision_b = other._get_file_revision(file_id, vf, 'other:')
449
last_revision_base = None
451
last_revision_base = base._get_file_revision(file_id, vf, 'base:')
452
return vf, last_revision_a, last_revision_b, last_revision_base
454
def plan_file_merge(self, file_id, other, base=None):
291
455
"""Generate a merge plan based on annotations.
293
457
If the file contains uncommitted changes in this tree, they will be
295
459
uncommitted changes in the other tree, they will be assigned to the
296
460
'other:' pseudo-revision.
298
from bzrlib import merge
299
annotated_a = list(self.annotate_iter(file_id,
300
_mod_revision.CURRENT_REVISION))
301
annotated_b = list(other.annotate_iter(file_id, 'other:'))
302
ancestors_a = self._get_ancestors(_mod_revision.CURRENT_REVISION)
303
ancestors_b = other._get_ancestors('other:')
304
return merge._plan_annotate_merge(annotated_a, annotated_b,
305
ancestors_a, ancestors_b)
462
data = self._get_plan_merge_data(file_id, other, base)
463
vf, last_revision_a, last_revision_b, last_revision_base = data
464
return vf.plan_merge(last_revision_a, last_revision_b,
467
def plan_file_lca_merge(self, file_id, other, base=None):
468
"""Generate a merge plan based lca-newness.
470
If the file contains uncommitted changes in this tree, they will be
471
attributed to the 'current:' pseudo-revision. If the file contains
472
uncommitted changes in the other tree, they will be assigned to the
473
'other:' pseudo-revision.
475
data = self._get_plan_merge_data(file_id, other, base)
476
vf, last_revision_a, last_revision_b, last_revision_base = data
477
return vf.plan_lca_merge(last_revision_a, last_revision_b,
480
def _iter_parent_trees(self):
481
"""Iterate through parent trees, defaulting to Tree.revision_tree."""
482
for revision_id in self.get_parent_ids():
484
yield self.revision_tree(revision_id)
485
except errors.NoSuchRevisionInTree:
486
yield self.repository.revision_tree(revision_id)
489
def _file_revision(revision_tree, file_id):
490
"""Determine the revision associated with a file in a given tree."""
491
revision_tree.lock_read()
493
return revision_tree.inventory[file_id].revision
495
revision_tree.unlock()
497
def _get_file_revision(self, file_id, vf, tree_revision):
498
"""Ensure that file_id, tree_revision is in vf to plan the merge."""
500
if getattr(self, '_repository', None) is None:
501
last_revision = tree_revision
502
parent_keys = [(file_id, self._file_revision(t, file_id)) for t in
503
self._iter_parent_trees()]
504
vf.add_lines((file_id, last_revision), parent_keys,
505
self.get_file(file_id).readlines())
506
repo = self.branch.repository
509
last_revision = self._file_revision(self, file_id)
510
base_vf = self._repository.texts
511
if base_vf not in vf.fallback_versionedfiles:
512
vf.fallback_versionedfiles.append(base_vf)
307
515
inventory = property(_get_inventory,
308
516
doc="Inventory of this Tree")
310
518
def _check_retrieved(self, ie, f):
311
519
if not __debug__:
313
521
fp = fingerprint_file(f)
316
524
if ie.text_size is not None:
317
525
if ie.text_size != fp['size']:
318
526
raise BzrError("mismatched size for file %r in %r" % (ie.file_id, self._store),
428
636
raise NotImplementedError(self.walkdirs)
431
class EmptyTree(Tree):
434
self._inventory = Inventory(root_id=None)
435
symbol_versioning.warn('EmptyTree is deprecated as of bzr 0.9 please'
436
' use repository.revision_tree instead.',
437
DeprecationWarning, stacklevel=2)
439
def get_parent_ids(self):
442
def get_symlink_target(self, file_id):
445
def has_filename(self, filename):
638
def supports_content_filtering(self):
448
def kind(self, file_id):
449
assert self._inventory[file_id].kind == "directory"
452
def list_files(self, include_root=False):
455
def __contains__(self, file_id):
456
return (file_id in self._inventory)
458
def get_file_sha1(self, file_id, path=None, stat_value=None):
641
def _content_filter_stack(self, path=None, file_id=None):
642
"""The stack of content filters for a path if filtering is supported.
644
Readers will be applied in first-to-last order.
645
Writers will be applied in last-to-first order.
646
Either the path or the file-id needs to be provided.
648
:param path: path relative to the root of the tree
650
:param file_id: file_id or None if unknown
651
:return: the list of filters - [] if there are none
653
filter_pref_names = filters._get_registered_names()
654
if len(filter_pref_names) == 0:
657
path = self.id2path(file_id)
658
prefs = self.iter_search_rules([path], filter_pref_names).next()
659
stk = filters._get_filter_stack_for(prefs)
660
if 'filters' in debug.debug_flags:
661
note("*** %s content-filter: %s => %r" % (path,prefs,stk))
664
def _content_filter_stack_provider(self):
665
"""A function that returns a stack of ContentFilters.
667
The function takes a path (relative to the top of the tree) and a
668
file-id as parameters.
670
:return: None if content filtering is not supported by this tree.
672
if self.supports_content_filtering():
673
return lambda path, file_id: \
674
self._content_filter_stack(path, file_id)
678
def iter_search_rules(self, path_names, pref_names=None,
679
_default_searcher=None):
680
"""Find the preferences for filenames in a tree.
682
:param path_names: an iterable of paths to find attributes for.
683
Paths are given relative to the root of the tree.
684
:param pref_names: the list of preferences to lookup - None for all
685
:param _default_searcher: private parameter to assist testing - don't use
686
:return: an iterator of tuple sequences, one per path-name.
687
See _RulesSearcher.get_items for details on the tuple sequence.
689
if _default_searcher is None:
690
_default_searcher = rules._per_user_searcher
691
searcher = self._get_rules_searcher(_default_searcher)
692
if searcher is not None:
693
if pref_names is not None:
694
for path in path_names:
695
yield searcher.get_selected_items(path, pref_names)
697
for path in path_names:
698
yield searcher.get_items(path)
701
def _get_rules_searcher(self, default_searcher):
702
"""Get the RulesSearcher for this tree given the default one."""
703
searcher = default_searcher
462
707
######################################################################
857
def _changes_from_entries(self, source_entry, target_entry,
858
source_path=None, target_path=None):
859
"""Generate a iter_changes tuple between source_entry and target_entry.
861
:param source_entry: An inventory entry from self.source, or None.
862
:param target_entry: An inventory entry from self.target, or None.
863
:param source_path: The path of source_entry, if known. If not known
864
it will be looked up.
865
:param target_path: The path of target_entry, if known. If not known
866
it will be looked up.
867
:return: A tuple, item 0 of which is an iter_changes result tuple, and
868
item 1 is True if there are any changes in the result tuple.
870
if source_entry is None:
871
if target_entry is None:
873
file_id = target_entry.file_id
875
file_id = source_entry.file_id
876
if source_entry is not None:
877
source_versioned = True
878
source_name = source_entry.name
879
source_parent = source_entry.parent_id
880
if source_path is None:
881
source_path = self.source.id2path(file_id)
882
source_kind, source_executable, source_stat = \
883
self.source._comparison_data(source_entry, source_path)
885
source_versioned = False
889
source_executable = None
890
if target_entry is not None:
891
target_versioned = True
892
target_name = target_entry.name
893
target_parent = target_entry.parent_id
894
if target_path is None:
895
target_path = self.target.id2path(file_id)
896
target_kind, target_executable, target_stat = \
897
self.target._comparison_data(target_entry, target_path)
899
target_versioned = False
903
target_executable = None
904
versioned = (source_versioned, target_versioned)
905
kind = (source_kind, target_kind)
906
changed_content = False
907
if source_kind != target_kind:
908
changed_content = True
909
elif source_kind == 'file':
910
if (self.source.get_file_sha1(file_id, source_path, source_stat) !=
911
self.target.get_file_sha1(file_id, target_path, target_stat)):
912
changed_content = True
913
elif source_kind == 'symlink':
914
if (self.source.get_symlink_target(file_id) !=
915
self.target.get_symlink_target(file_id)):
916
changed_content = True
917
# XXX: Yes, the indentation below is wrong. But fixing it broke
918
# test_merge.TestMergerEntriesLCAOnDisk.
919
# test_nested_tree_subtree_renamed_and_modified. We'll wait for
920
# the fix from bzr.dev -- vila 2009026
921
elif source_kind == 'tree-reference':
922
if (self.source.get_reference_revision(file_id, source_path)
923
!= self.target.get_reference_revision(file_id, target_path)):
924
changed_content = True
925
parent = (source_parent, target_parent)
926
name = (source_name, target_name)
927
executable = (source_executable, target_executable)
928
if (changed_content is not False or versioned[0] != versioned[1]
929
or parent[0] != parent[1] or name[0] != name[1] or
930
executable[0] != executable[1]):
934
return (file_id, (source_path, target_path), changed_content,
935
versioned, parent, name, kind, executable), changes
615
938
def compare(self, want_unchanged=False, specific_files=None,
616
939
extra_trees=None, require_versioned=False, include_root=False,
702
1042
all_unversioned = deque()
704
from_entries_by_dir = list(self.source.inventory.iter_entries_by_dir(
1044
from_entries_by_dir = list(self.source.iter_entries_by_dir(
705
1045
specific_file_ids=specific_file_ids))
706
1046
from_data = dict((e.file_id, (p, e)) for p, e in from_entries_by_dir)
707
to_entries_by_dir = list(self.target.inventory.iter_entries_by_dir(
1047
to_entries_by_dir = list(self.target.iter_entries_by_dir(
708
1048
specific_file_ids=specific_file_ids))
709
1049
num_entries = len(from_entries_by_dir) + len(to_entries_by_dir)
711
# the unversioned path lookup only occurs on real trees - where there
1051
# the unversioned path lookup only occurs on real trees - where there
712
1052
# can be extras. So the fake_entry is solely used to look up
713
1053
# executable it values when execute is not supported.
714
1054
fake_entry = InventoryFile('unused', 'unused', 'unused')
715
for to_path, to_entry in to_entries_by_dir:
716
while all_unversioned and all_unversioned[0][0] < to_path.split('/'):
1055
for target_path, target_entry in to_entries_by_dir:
1056
while (all_unversioned and
1057
all_unversioned[0][0] < target_path.split('/')):
717
1058
unversioned_path = all_unversioned.popleft()
718
to_kind, to_executable, to_stat = \
1059
target_kind, target_executable, target_stat = \
719
1060
self.target._comparison_data(fake_entry, unversioned_path[1])
720
1061
yield (None, (None, unversioned_path[1]), True, (False, False),
722
1063
(None, unversioned_path[0][-1]),
724
(None, to_executable))
725
file_id = to_entry.file_id
726
to_paths[file_id] = to_path
1064
(None, target_kind),
1065
(None, target_executable))
1066
source_path, source_entry = from_data.get(target_entry.file_id,
1068
result, changes = self._changes_from_entries(source_entry,
1069
target_entry, source_path=source_path, target_path=target_path)
1070
to_paths[result[0]] = result[1][1]
727
1071
entry_count += 1
728
changed_content = False
729
from_path, from_entry = from_data.get(file_id, (None, None))
730
from_versioned = (from_entry is not None)
731
if from_entry is not None:
732
from_versioned = True
733
from_name = from_entry.name
734
from_parent = from_entry.parent_id
735
from_kind, from_executable, from_stat = \
736
self.source._comparison_data(from_entry, from_path)
737
1073
entry_count += 1
739
from_versioned = False
743
from_executable = None
744
versioned = (from_versioned, True)
745
to_kind, to_executable, to_stat = \
746
self.target._comparison_data(to_entry, to_path)
747
kind = (from_kind, to_kind)
748
if kind[0] != kind[1]:
749
changed_content = True
750
elif from_kind == 'file':
751
from_size = self.source._file_size(from_entry, from_stat)
752
to_size = self.target._file_size(to_entry, to_stat)
753
if from_size != to_size:
754
changed_content = True
755
elif (self.source.get_file_sha1(file_id, from_path, from_stat) !=
756
self.target.get_file_sha1(file_id, to_path, to_stat)):
757
changed_content = True
758
elif from_kind == 'symlink':
759
if (self.source.get_symlink_target(file_id) !=
760
self.target.get_symlink_target(file_id)):
761
changed_content = True
762
elif from_kind == 'tree-reference':
763
if (self.source.get_reference_revision(file_id, from_path)
764
!= self.target.get_reference_revision(file_id, to_path)):
765
changed_content = True
766
parent = (from_parent, to_entry.parent_id)
767
name = (from_name, to_entry.name)
768
executable = (from_executable, to_executable)
769
1074
if pb is not None:
770
1075
pb.update('comparing files', entry_count, num_entries)
771
if (changed_content is not False or versioned[0] != versioned[1]
772
or parent[0] != parent[1] or name[0] != name[1] or
773
executable[0] != executable[1] or include_unchanged):
774
yield (file_id, (from_path, to_path), changed_content,
775
versioned, parent, name, kind, executable)
1076
if changes or include_unchanged:
1077
if specific_file_ids is not None:
1078
new_parent_id = result[4][1]
1079
precise_file_ids.add(new_parent_id)
1080
changed_file_ids.append(result[0])
1082
# Ensure correct behaviour for reparented/added specific files.
1083
if specific_files is not None:
1084
# Record output dirs
1085
if result[6][1] == 'directory':
1086
seen_dirs.add(result[0])
1087
# Record parents of reparented/added entries.
1088
versioned = result[3]
1090
if not versioned[0] or parents[0] != parents[1]:
1091
seen_parents.add(parents[1])
777
1092
while all_unversioned:
778
1093
# yield any trailing unversioned paths
779
1094
unversioned_path = all_unversioned.popleft()
818
1121
self.source._comparison_data(from_entry, path)
819
1122
kind = (from_kind, None)
820
1123
executable = (from_executable, None)
821
changed_content = True
1124
changed_content = from_kind is not None
822
1125
# the parent's path is necessarily known at this point.
1126
changed_file_ids.append(file_id)
823
1127
yield(file_id, (path, to_path), changed_content, versioned, parent,
824
1128
name, kind, executable)
827
# This was deprecated before 0.12, but did not have an official warning
828
@symbol_versioning.deprecated_function(symbol_versioning.zero_twelve)
829
def RevisionTree(*args, **kwargs):
830
"""RevisionTree has moved to bzrlib.revisiontree.RevisionTree()
832
Accessing it as bzrlib.tree.RevisionTree has been deprecated as of
835
from bzrlib.revisiontree import RevisionTree as _RevisionTree
836
return _RevisionTree(*args, **kwargs)
1129
changed_file_ids = set(changed_file_ids)
1130
if specific_file_ids is not None:
1131
for result in self._handle_precise_ids(precise_file_ids,
1135
def _get_entry(self, tree, file_id):
1136
"""Get an inventory entry from a tree, with missing entries as None.
1138
If the tree raises NotImplementedError on accessing .inventory, then
1139
this is worked around using iter_entries_by_dir on just the file id
1142
:param tree: The tree to lookup the entry in.
1143
:param file_id: The file_id to lookup.
1146
inventory = tree.inventory
1147
except NotImplementedError:
1148
# No inventory available.
1150
iterator = tree.iter_entries_by_dir(specific_file_ids=[file_id])
1151
return iterator.next()[1]
1152
except StopIteration:
1156
return inventory[file_id]
1157
except errors.NoSuchId:
1160
def _handle_precise_ids(self, precise_file_ids, changed_file_ids,
1161
discarded_changes=None):
1162
"""Fill out a partial iter_changes to be consistent.
1164
:param precise_file_ids: The file ids of parents that were seen during
1166
:param changed_file_ids: The file ids of already emitted items.
1167
:param discarded_changes: An optional dict of precalculated
1168
iter_changes items which the partial iter_changes had not output
1170
:return: A generator of iter_changes items to output.
1172
# process parents of things that had changed under the users
1173
# requested paths to prevent incorrect paths or parent ids which
1174
# aren't in the tree.
1175
while precise_file_ids:
1176
precise_file_ids.discard(None)
1177
# Don't emit file_ids twice
1178
precise_file_ids.difference_update(changed_file_ids)
1179
if not precise_file_ids:
1181
# If the there was something at a given output path in source, we
1182
# have to include the entry from source in the delta, or we would
1183
# be putting this entry into a used path.
1185
for parent_id in precise_file_ids:
1187
paths.append(self.target.id2path(parent_id))
1188
except errors.NoSuchId:
1189
# This id has been dragged in from the source by delta
1190
# expansion and isn't present in target at all: we don't
1191
# need to check for path collisions on it.
1194
old_id = self.source.path2id(path)
1195
precise_file_ids.add(old_id)
1196
precise_file_ids.discard(None)
1197
current_ids = precise_file_ids
1198
precise_file_ids = set()
1199
# We have to emit all of precise_file_ids that have been altered.
1200
# We may have to output the children of some of those ids if any
1201
# directories have stopped being directories.
1202
for file_id in current_ids:
1204
if discarded_changes:
1205
result = discarded_changes.get(file_id)
1210
old_entry = self._get_entry(self.source, file_id)
1211
new_entry = self._get_entry(self.target, file_id)
1212
result, changes = self._changes_from_entries(
1213
old_entry, new_entry)
1216
# Get this parents parent to examine.
1217
new_parent_id = result[4][1]
1218
precise_file_ids.add(new_parent_id)
1220
if (result[6][0] == 'directory' and
1221
result[6][1] != 'directory'):
1222
# This stopped being a directory, the old children have
1224
if old_entry is None:
1225
# Reusing a discarded change.
1226
old_entry = self._get_entry(self.source, file_id)
1227
for child in old_entry.children.values():
1228
precise_file_ids.add(child.file_id)
1229
changed_file_ids.add(result[0])
1233
class MultiWalker(object):
1234
"""Walk multiple trees simultaneously, getting combined results."""
1236
# Note: This could be written to not assume you can do out-of-order
1237
# lookups. Instead any nodes that don't match in all trees could be
1238
# marked as 'deferred', and then returned in the final cleanup loop.
1239
# For now, I think it is "nicer" to return things as close to the
1240
# "master_tree" order as we can.
1242
def __init__(self, master_tree, other_trees):
1243
"""Create a new MultiWalker.
1245
All trees being walked must implement "iter_entries_by_dir()", such
1246
that they yield (path, object) tuples, where that object will have a
1247
'.file_id' member, that can be used to check equality.
1249
:param master_tree: All trees will be 'slaved' to the master_tree such
1250
that nodes in master_tree will be used as 'first-pass' sync points.
1251
Any nodes that aren't in master_tree will be merged in a second
1253
:param other_trees: A list of other trees to walk simultaneously.
1255
self._master_tree = master_tree
1256
self._other_trees = other_trees
1258
# Keep track of any nodes that were properly processed just out of
1259
# order, that way we don't return them at the end, we don't have to
1260
# track *all* processed file_ids, just the out-of-order ones
1261
self._out_of_order_processed = set()
1264
def _step_one(iterator):
1265
"""Step an iter_entries_by_dir iterator.
1267
:return: (has_more, path, ie)
1268
If has_more is False, path and ie will be None.
1271
path, ie = iterator.next()
1272
except StopIteration:
1273
return False, None, None
1275
return True, path, ie
1278
def _cmp_path_by_dirblock(path1, path2):
1279
"""Compare two paths based on what directory they are in.
1281
This generates a sort order, such that all children of a directory are
1282
sorted together, and grandchildren are in the same order as the
1283
children appear. But all grandchildren come after all children.
1285
:param path1: first path
1286
:param path2: the second path
1287
:return: negative number if ``path1`` comes first,
1288
0 if paths are equal
1289
and a positive number if ``path2`` sorts first
1291
# Shortcut this special case
1294
# This is stolen from _dirstate_helpers_py.py, only switching it to
1295
# Unicode objects. Consider using encode_utf8() and then using the
1296
# optimized versions, or maybe writing optimized unicode versions.
1297
if not isinstance(path1, unicode):
1298
raise TypeError("'path1' must be a unicode string, not %s: %r"
1299
% (type(path1), path1))
1300
if not isinstance(path2, unicode):
1301
raise TypeError("'path2' must be a unicode string, not %s: %r"
1302
% (type(path2), path2))
1303
return cmp(MultiWalker._path_to_key(path1),
1304
MultiWalker._path_to_key(path2))
1307
def _path_to_key(path):
1308
dirname, basename = osutils.split(path)
1309
return (dirname.split(u'/'), basename)
1311
def _lookup_by_file_id(self, extra_entries, other_tree, file_id):
1312
"""Lookup an inventory entry by file_id.
1314
This is called when an entry is missing in the normal order.
1315
Generally this is because a file was either renamed, or it was
1316
deleted/added. If the entry was found in the inventory and not in
1317
extra_entries, it will be added to self._out_of_order_processed
1319
:param extra_entries: A dictionary of {file_id: (path, ie)}. This
1320
should be filled with entries that were found before they were
1321
used. If file_id is present, it will be removed from the
1323
:param other_tree: The Tree to search, in case we didn't find the entry
1325
:param file_id: The file_id to look for
1326
:return: (path, ie) if found or (None, None) if not present.
1328
if file_id in extra_entries:
1329
return extra_entries.pop(file_id)
1330
# TODO: Is id2path better as the first call, or is
1331
# inventory[file_id] better as a first check?
1333
cur_path = other_tree.id2path(file_id)
1334
except errors.NoSuchId:
1336
if cur_path is None:
1339
self._out_of_order_processed.add(file_id)
1340
cur_ie = other_tree.inventory[file_id]
1341
return (cur_path, cur_ie)
1344
"""Match up the values in the different trees."""
1345
for result in self._walk_master_tree():
1347
self._finish_others()
1348
for result in self._walk_others():
1351
def _walk_master_tree(self):
1352
"""First pass, walk all trees in lock-step.
1354
When we are done, all nodes in the master_tree will have been
1355
processed. _other_walkers, _other_entries, and _others_extra will be
1356
set on 'self' for future processing.
1358
# This iterator has the most "inlining" done, because it tends to touch
1359
# every file in the tree, while the others only hit nodes that don't
1361
master_iterator = self._master_tree.iter_entries_by_dir()
1363
other_walkers = [other.iter_entries_by_dir()
1364
for other in self._other_trees]
1365
other_entries = [self._step_one(walker) for walker in other_walkers]
1366
# Track extra nodes in the other trees
1367
others_extra = [{} for i in xrange(len(self._other_trees))]
1369
master_has_more = True
1370
step_one = self._step_one
1371
lookup_by_file_id = self._lookup_by_file_id
1372
out_of_order_processed = self._out_of_order_processed
1374
while master_has_more:
1375
(master_has_more, path, master_ie) = step_one(master_iterator)
1376
if not master_has_more:
1379
file_id = master_ie.file_id
1381
other_values_append = other_values.append
1382
next_other_entries = []
1383
next_other_entries_append = next_other_entries.append
1384
for idx, (other_has_more, other_path, other_ie) in enumerate(other_entries):
1385
if not other_has_more:
1386
other_values_append(lookup_by_file_id(
1387
others_extra[idx], self._other_trees[idx], file_id))
1388
next_other_entries_append((False, None, None))
1389
elif file_id == other_ie.file_id:
1390
# This is the critical code path, as most of the entries
1391
# should match between most trees.
1392
other_values_append((other_path, other_ie))
1393
next_other_entries_append(step_one(other_walkers[idx]))
1395
# This walker did not match, step it until it either
1396
# matches, or we know we are past the current walker.
1397
other_walker = other_walkers[idx]
1398
other_extra = others_extra[idx]
1399
while (other_has_more and
1400
self._cmp_path_by_dirblock(other_path, path) < 0):
1401
other_file_id = other_ie.file_id
1402
if other_file_id not in out_of_order_processed:
1403
other_extra[other_file_id] = (other_path, other_ie)
1404
other_has_more, other_path, other_ie = \
1405
step_one(other_walker)
1406
if other_has_more and other_ie.file_id == file_id:
1407
# We ended up walking to this point, match and step
1409
other_values_append((other_path, other_ie))
1410
other_has_more, other_path, other_ie = \
1411
step_one(other_walker)
1413
# This record isn't in the normal order, see if it
1415
other_values_append(lookup_by_file_id(
1416
other_extra, self._other_trees[idx], file_id))
1417
next_other_entries_append((other_has_more, other_path,
1419
other_entries = next_other_entries
1421
# We've matched all the walkers, yield this datapoint
1422
yield path, file_id, master_ie, other_values
1423
self._other_walkers = other_walkers
1424
self._other_entries = other_entries
1425
self._others_extra = others_extra
1427
def _finish_others(self):
1428
"""Finish walking the other iterators, so we get all entries."""
1429
for idx, info in enumerate(self._other_entries):
1430
other_extra = self._others_extra[idx]
1431
(other_has_more, other_path, other_ie) = info
1432
while other_has_more:
1433
other_file_id = other_ie.file_id
1434
if other_file_id not in self._out_of_order_processed:
1435
other_extra[other_file_id] = (other_path, other_ie)
1436
other_has_more, other_path, other_ie = \
1437
self._step_one(self._other_walkers[idx])
1438
del self._other_entries
1440
def _walk_others(self):
1441
"""Finish up by walking all the 'deferred' nodes."""
1442
# TODO: One alternative would be to grab all possible unprocessed
1443
# file_ids, and then sort by path, and then yield them. That
1444
# might ensure better ordering, in case a caller strictly
1445
# requires parents before children.
1446
for idx, other_extra in enumerate(self._others_extra):
1447
others = sorted(other_extra.itervalues(),
1448
key=lambda x: self._path_to_key(x[0]))
1449
for other_path, other_ie in others:
1450
file_id = other_ie.file_id
1451
# We don't need to check out_of_order_processed here, because
1452
# the lookup_by_file_id will be removing anything processed
1453
# from the extras cache
1454
other_extra.pop(file_id)
1455
other_values = [(None, None) for i in xrange(idx)]
1456
other_values.append((other_path, other_ie))
1457
for alt_idx, alt_extra in enumerate(self._others_extra[idx+1:]):
1458
alt_idx = alt_idx + idx + 1
1459
alt_extra = self._others_extra[alt_idx]
1460
alt_tree = self._other_trees[alt_idx]
1461
other_values.append(self._lookup_by_file_id(
1462
alt_extra, alt_tree, file_id))
1463
yield other_path, file_id, None, other_values