357
400
return vf.plan_lca_merge(last_revision_a, last_revision_b,
358
401
last_revision_base)
403
def _iter_parent_trees(self):
404
"""Iterate through parent trees, defaulting to Tree.revision_tree."""
405
for revision_id in self.get_parent_ids():
407
yield self.revision_tree(revision_id)
408
except errors.NoSuchRevisionInTree:
409
yield self.repository.revision_tree(revision_id)
412
def _file_revision(revision_tree, file_id):
413
"""Determine the revision associated with a file in a given tree."""
414
revision_tree.lock_read()
416
return revision_tree.inventory[file_id].revision
418
revision_tree.unlock()
360
420
def _get_file_revision(self, file_id, vf, tree_revision):
361
def file_revision(revision_tree):
362
revision_tree.lock_read()
364
return revision_tree.inventory[file_id].revision
366
revision_tree.unlock()
368
def iter_parent_trees():
369
for revision_id in self.get_parent_ids():
371
yield self.revision_tree(revision_id)
373
yield self.repository.revision_tree(revision_id)
375
if getattr(self, '_get_weave', None) is None:
421
"""Ensure that file_id, tree_revision is in vf to plan the merge."""
423
if getattr(self, '_repository', None) is None:
376
424
last_revision = tree_revision
377
parent_revisions = [file_revision(t) for t in iter_parent_trees()]
378
vf.add_lines(last_revision, parent_revisions,
425
parent_keys = [(file_id, self._file_revision(t, file_id)) for t in
426
self._iter_parent_trees()]
427
vf.add_lines((file_id, last_revision), parent_keys,
379
428
self.get_file(file_id).readlines())
380
429
repo = self.branch.repository
381
transaction = repo.get_transaction()
382
base_vf = repo.weave_store.get_weave(file_id, transaction)
384
last_revision = file_revision(self)
385
base_vf = self._get_weave(file_id)
386
vf.fallback_versionedfiles.append(base_vf)
432
last_revision = self._file_revision(self, file_id)
433
base_vf = self._repository.texts
434
if base_vf not in vf.fallback_versionedfiles:
435
vf.fallback_versionedfiles.append(base_vf)
387
436
return last_revision
389
438
inventory = property(_get_inventory,
898
978
self.source._comparison_data(from_entry, path)
899
979
kind = (from_kind, None)
900
980
executable = (from_executable, None)
901
changed_content = True
981
changed_content = from_kind is not None
902
982
# the parent's path is necessarily known at this point.
903
983
yield(file_id, (path, to_path), changed_content, versioned, parent,
904
984
name, kind, executable)
987
class MultiWalker(object):
988
"""Walk multiple trees simultaneously, getting combined results."""
990
# Note: This could be written to not assume you can do out-of-order
991
# lookups. Instead any nodes that don't match in all trees could be
992
# marked as 'deferred', and then returned in the final cleanup loop.
993
# For now, I think it is "nicer" to return things as close to the
994
# "master_tree" order as we can.
996
def __init__(self, master_tree, other_trees):
997
"""Create a new MultiWalker.
999
All trees being walked must implement "iter_entries_by_dir()", such
1000
that they yield (path, object) tuples, where that object will have a
1001
'.file_id' member, that can be used to check equality.
1003
:param master_tree: All trees will be 'slaved' to the master_tree such
1004
that nodes in master_tree will be used as 'first-pass' sync points.
1005
Any nodes that aren't in master_tree will be merged in a second
1007
:param other_trees: A list of other trees to walk simultaneously.
1009
self._master_tree = master_tree
1010
self._other_trees = other_trees
1012
# Keep track of any nodes that were properly processed just out of
1013
# order, that way we don't return them at the end, we don't have to
1014
# track *all* processed file_ids, just the out-of-order ones
1015
self._out_of_order_processed = set()
1018
def _step_one(iterator):
1019
"""Step an iter_entries_by_dir iterator.
1021
:return: (has_more, path, ie)
1022
If has_more is False, path and ie will be None.
1025
path, ie = iterator.next()
1026
except StopIteration:
1027
return False, None, None
1029
return True, path, ie
1032
def _cmp_path_by_dirblock(path1, path2):
1033
"""Compare two paths based on what directory they are in.
1035
This generates a sort order, such that all children of a directory are
1036
sorted together, and grandchildren are in the same order as the
1037
children appear. But all grandchildren come after all children.
1039
:param path1: first path
1040
:param path2: the second path
1041
:return: negative number if ``path1`` comes first,
1042
0 if paths are equal
1043
and a positive number if ``path2`` sorts first
1045
# Shortcut this special case
1048
# This is stolen from _dirstate_helpers_py.py, only switching it to
1049
# Unicode objects. Consider using encode_utf8() and then using the
1050
# optimized versions, or maybe writing optimized unicode versions.
1051
if not isinstance(path1, unicode):
1052
raise TypeError("'path1' must be a unicode string, not %s: %r"
1053
% (type(path1), path1))
1054
if not isinstance(path2, unicode):
1055
raise TypeError("'path2' must be a unicode string, not %s: %r"
1056
% (type(path2), path2))
1057
return cmp(MultiWalker._path_to_key(path1),
1058
MultiWalker._path_to_key(path2))
1061
def _path_to_key(path):
1062
dirname, basename = osutils.split(path)
1063
return (dirname.split(u'/'), basename)
1065
def _lookup_by_file_id(self, extra_entries, other_tree, file_id):
1066
"""Lookup an inventory entry by file_id.
1068
This is called when an entry is missing in the normal order.
1069
Generally this is because a file was either renamed, or it was
1070
deleted/added. If the entry was found in the inventory and not in
1071
extra_entries, it will be added to self._out_of_order_processed
1073
:param extra_entries: A dictionary of {file_id: (path, ie)}. This
1074
should be filled with entries that were found before they were
1075
used. If file_id is present, it will be removed from the
1077
:param other_tree: The Tree to search, in case we didn't find the entry
1079
:param file_id: The file_id to look for
1080
:return: (path, ie) if found or (None, None) if not present.
1082
if file_id in extra_entries:
1083
return extra_entries.pop(file_id)
1084
# TODO: Is id2path better as the first call, or is
1085
# inventory[file_id] better as a first check?
1087
cur_path = other_tree.id2path(file_id)
1088
except errors.NoSuchId:
1090
if cur_path is None:
1093
self._out_of_order_processed.add(file_id)
1094
cur_ie = other_tree.inventory[file_id]
1095
return (cur_path, cur_ie)
1098
"""Match up the values in the different trees."""
1099
for result in self._walk_master_tree():
1101
self._finish_others()
1102
for result in self._walk_others():
1105
def _walk_master_tree(self):
1106
"""First pass, walk all trees in lock-step.
1108
When we are done, all nodes in the master_tree will have been
1109
processed. _other_walkers, _other_entries, and _others_extra will be
1110
set on 'self' for future processing.
1112
# This iterator has the most "inlining" done, because it tends to touch
1113
# every file in the tree, while the others only hit nodes that don't
1115
master_iterator = self._master_tree.iter_entries_by_dir()
1117
other_walkers = [other.iter_entries_by_dir()
1118
for other in self._other_trees]
1119
other_entries = [self._step_one(walker) for walker in other_walkers]
1120
# Track extra nodes in the other trees
1121
others_extra = [{} for i in xrange(len(self._other_trees))]
1123
master_has_more = True
1124
step_one = self._step_one
1125
lookup_by_file_id = self._lookup_by_file_id
1126
out_of_order_processed = self._out_of_order_processed
1128
while master_has_more:
1129
(master_has_more, path, master_ie) = step_one(master_iterator)
1130
if not master_has_more:
1133
file_id = master_ie.file_id
1135
other_values_append = other_values.append
1136
next_other_entries = []
1137
next_other_entries_append = next_other_entries.append
1138
for idx, (other_has_more, other_path, other_ie) in enumerate(other_entries):
1139
if not other_has_more:
1140
other_values_append(lookup_by_file_id(
1141
others_extra[idx], self._other_trees[idx], file_id))
1142
next_other_entries_append((False, None, None))
1143
elif file_id == other_ie.file_id:
1144
# This is the critical code path, as most of the entries
1145
# should match between most trees.
1146
other_values_append((other_path, other_ie))
1147
next_other_entries_append(step_one(other_walkers[idx]))
1149
# This walker did not match, step it until it either
1150
# matches, or we know we are past the current walker.
1151
other_walker = other_walkers[idx]
1152
other_extra = others_extra[idx]
1153
while (other_has_more and
1154
self._cmp_path_by_dirblock(other_path, path) < 0):
1155
other_file_id = other_ie.file_id
1156
if other_file_id not in out_of_order_processed:
1157
other_extra[other_file_id] = (other_path, other_ie)
1158
other_has_more, other_path, other_ie = \
1159
step_one(other_walker)
1160
if other_has_more and other_ie.file_id == file_id:
1161
# We ended up walking to this point, match and step
1163
other_values_append((other_path, other_ie))
1164
other_has_more, other_path, other_ie = \
1165
step_one(other_walker)
1167
# This record isn't in the normal order, see if it
1169
other_values_append(lookup_by_file_id(
1170
other_extra, self._other_trees[idx], file_id))
1171
next_other_entries_append((other_has_more, other_path,
1173
other_entries = next_other_entries
1175
# We've matched all the walkers, yield this datapoint
1176
yield path, file_id, master_ie, other_values
1177
self._other_walkers = other_walkers
1178
self._other_entries = other_entries
1179
self._others_extra = others_extra
1181
def _finish_others(self):
1182
"""Finish walking the other iterators, so we get all entries."""
1183
for idx, info in enumerate(self._other_entries):
1184
other_extra = self._others_extra[idx]
1185
(other_has_more, other_path, other_ie) = info
1186
while other_has_more:
1187
other_file_id = other_ie.file_id
1188
if other_file_id not in self._out_of_order_processed:
1189
other_extra[other_file_id] = (other_path, other_ie)
1190
other_has_more, other_path, other_ie = \
1191
self._step_one(self._other_walkers[idx])
1192
del self._other_entries
1194
def _walk_others(self):
1195
"""Finish up by walking all the 'deferred' nodes."""
1196
# TODO: One alternative would be to grab all possible unprocessed
1197
# file_ids, and then sort by path, and then yield them. That
1198
# might ensure better ordering, in case a caller strictly
1199
# requires parents before children.
1200
for idx, other_extra in enumerate(self._others_extra):
1201
others = sorted(other_extra.itervalues(),
1202
key=lambda x: self._path_to_key(x[0]))
1203
for other_path, other_ie in others:
1204
file_id = other_ie.file_id
1205
# We don't need to check out_of_order_processed here, because
1206
# the lookup_by_file_id will be removing anything processed
1207
# from the extras cache
1208
other_extra.pop(file_id)
1209
other_values = [(None, None) for i in xrange(idx)]
1210
other_values.append((other_path, other_ie))
1211
for alt_idx, alt_extra in enumerate(self._others_extra[idx+1:]):
1212
alt_idx = alt_idx + idx + 1
1213
alt_extra = self._others_extra[alt_idx]
1214
alt_tree = self._other_trees[alt_idx]
1215
other_values.append(self._lookup_by_file_id(
1216
alt_extra, alt_tree, file_id))
1217
yield other_path, file_id, None, other_values
added
removed
bzrlib/tree.py
