957
952
self.source._comparison_data(from_entry, path)
958
953
kind = (from_kind, None)
959
954
executable = (from_executable, None)
960
changed_content = True
955
changed_content = from_kind is not None
961
956
# the parent's path is necessarily known at this point.
962
957
yield(file_id, (path, to_path), changed_content, versioned, parent,
963
958
name, kind, executable)
961
class MultiWalker(object):
962
"""Walk multiple trees simultaneously, getting combined results."""
964
# Note: This could be written to not assume you can do out-of-order
965
# lookups. Instead any nodes that don't match in all trees could be
966
# marked as 'deferred', and then returned in the final cleanup loop.
967
# For now, I think it is "nicer" to return things as close to the
968
# "master_tree" order as we can.
970
def __init__(self, master_tree, other_trees):
971
"""Create a new MultiWalker.
973
All trees being walked must implement "iter_entries_by_dir()", such
974
that they yield (path, object) tuples, where that object will have a
975
'.file_id' member, that can be used to check equality.
977
:param master_tree: All trees will be 'slaved' to the master_tree such
978
that nodes in master_tree will be used as 'first-pass' sync points.
979
Any nodes that aren't in master_tree will be merged in a second
981
:param other_trees: A list of other trees to walk simultaneously.
983
self._master_tree = master_tree
984
self._other_trees = other_trees
986
# Keep track of any nodes that were properly processed just out of
987
# order, that way we don't return them at the end, we don't have to
988
# track *all* processed file_ids, just the out-of-order ones
989
self._out_of_order_processed = set()
992
def _step_one(iterator):
993
"""Step an iter_entries_by_dir iterator.
995
:return: (has_more, path, ie)
996
If has_more is False, path and ie will be None.
999
path, ie = iterator.next()
1000
except StopIteration:
1001
return False, None, None
1003
return True, path, ie
1006
def _cmp_path_by_dirblock(path1, path2):
1007
"""Compare two paths based on what directory they are in.
1009
This generates a sort order, such that all children of a directory are
1010
sorted together, and grandchildren are in the same order as the
1011
children appear. But all grandchildren come after all children.
1013
:param path1: first path
1014
:param path2: the second path
1015
:return: negative number if ``path1`` comes first,
1016
0 if paths are equal
1017
and a positive number if ``path2`` sorts first
1019
# Shortcut this special case
1022
# This is stolen from _dirstate_helpers_py.py, only switching it to
1023
# Unicode objects. Consider using encode_utf8() and then using the
1024
# optimized versions, or maybe writing optimized unicode versions.
1025
if not isinstance(path1, unicode):
1026
raise TypeError("'path1' must be a unicode string, not %s: %r"
1027
% (type(path1), path1))
1028
if not isinstance(path2, unicode):
1029
raise TypeError("'path2' must be a unicode string, not %s: %r"
1030
% (type(path2), path2))
1031
return cmp(MultiWalker._path_to_key(path1),
1032
MultiWalker._path_to_key(path2))
1035
def _path_to_key(path):
1036
dirname, basename = osutils.split(path)
1037
return (dirname.split(u'/'), basename)
1039
def _lookup_by_file_id(self, extra_entries, other_tree, file_id):
1040
"""Lookup an inventory entry by file_id.
1042
This is called when an entry is missing in the normal order.
1043
Generally this is because a file was either renamed, or it was
1044
deleted/added. If the entry was found in the inventory and not in
1045
extra_entries, it will be added to self._out_of_order_processed
1047
:param extra_entries: A dictionary of {file_id: (path, ie)}. This
1048
should be filled with entries that were found before they were
1049
used. If file_id is present, it will be removed from the
1051
:param other_tree: The Tree to search, in case we didn't find the entry
1053
:param file_id: The file_id to look for
1054
:return: (path, ie) if found or (None, None) if not present.
1056
if file_id in extra_entries:
1057
return extra_entries.pop(file_id)
1058
# TODO: Is id2path better as the first call, or is
1059
# inventory[file_id] better as a first check?
1061
cur_path = other_tree.id2path(file_id)
1062
except errors.NoSuchId:
1064
if cur_path is None:
1067
self._out_of_order_processed.add(file_id)
1068
cur_ie = other_tree.inventory[file_id]
1069
return (cur_path, cur_ie)
1072
"""Match up the values in the different trees."""
1073
for result in self._walk_master_tree():
1075
self._finish_others()
1076
for result in self._walk_others():
1079
def _walk_master_tree(self):
1080
"""First pass, walk all trees in lock-step.
1082
When we are done, all nodes in the master_tree will have been
1083
processed. _other_walkers, _other_entries, and _others_extra will be
1084
set on 'self' for future processing.
1086
# This iterator has the most "inlining" done, because it tends to touch
1087
# every file in the tree, while the others only hit nodes that don't
1089
master_iterator = self._master_tree.iter_entries_by_dir()
1091
other_walkers = [other.iter_entries_by_dir()
1092
for other in self._other_trees]
1093
other_entries = [self._step_one(walker) for walker in other_walkers]
1094
# Track extra nodes in the other trees
1095
others_extra = [{} for i in xrange(len(self._other_trees))]
1097
master_has_more = True
1098
step_one = self._step_one
1099
lookup_by_file_id = self._lookup_by_file_id
1100
out_of_order_processed = self._out_of_order_processed
1102
while master_has_more:
1103
(master_has_more, path, master_ie) = step_one(master_iterator)
1104
if not master_has_more:
1107
file_id = master_ie.file_id
1109
other_values_append = other_values.append
1110
next_other_entries = []
1111
next_other_entries_append = next_other_entries.append
1112
for idx, (other_has_more, other_path, other_ie) in enumerate(other_entries):
1113
if not other_has_more:
1114
other_values_append(lookup_by_file_id(
1115
others_extra[idx], self._other_trees[idx], file_id))
1116
next_other_entries_append((False, None, None))
1117
elif file_id == other_ie.file_id:
1118
# This is the critical code path, as most of the entries
1119
# should match between most trees.
1120
other_values_append((other_path, other_ie))
1121
next_other_entries_append(step_one(other_walkers[idx]))
1123
# This walker did not match, step it until it either
1124
# matches, or we know we are past the current walker.
1125
other_walker = other_walkers[idx]
1126
other_extra = others_extra[idx]
1127
while (other_has_more and
1128
self._cmp_path_by_dirblock(other_path, path) < 0):
1129
other_file_id = other_ie.file_id
1130
if other_file_id not in out_of_order_processed:
1131
other_extra[other_file_id] = (other_path, other_ie)
1132
other_has_more, other_path, other_ie = \
1133
step_one(other_walker)
1134
if other_has_more and other_ie.file_id == file_id:
1135
# We ended up walking to this point, match and step
1137
other_values_append((other_path, other_ie))
1138
other_has_more, other_path, other_ie = \
1139
step_one(other_walker)
1141
# This record isn't in the normal order, see if it
1143
other_values_append(lookup_by_file_id(
1144
other_extra, self._other_trees[idx], file_id))
1145
next_other_entries_append((other_has_more, other_path,
1147
other_entries = next_other_entries
1149
# We've matched all the walkers, yield this datapoint
1150
yield path, file_id, master_ie, other_values
1151
self._other_walkers = other_walkers
1152
self._other_entries = other_entries
1153
self._others_extra = others_extra
1155
def _finish_others(self):
1156
"""Finish walking the other iterators, so we get all entries."""
1157
for idx, info in enumerate(self._other_entries):
1158
other_extra = self._others_extra[idx]
1159
(other_has_more, other_path, other_ie) = info
1160
while other_has_more:
1161
other_file_id = other_ie.file_id
1162
if other_file_id not in self._out_of_order_processed:
1163
other_extra[other_file_id] = (other_path, other_ie)
1164
other_has_more, other_path, other_ie = \
1165
self._step_one(self._other_walkers[idx])
1166
del self._other_entries
1168
def _walk_others(self):
1169
"""Finish up by walking all the 'deferred' nodes."""
1170
# TODO: One alternative would be to grab all possible unprocessed
1171
# file_ids, and then sort by path, and then yield them. That
1172
# might ensure better ordering, in case a caller strictly
1173
# requires parents before children.
1174
for idx, other_extra in enumerate(self._others_extra):
1175
others = sorted(other_extra.itervalues(),
1176
key=lambda x: self._path_to_key(x[0]))
1177
for other_path, other_ie in others:
1178
file_id = other_ie.file_id
1179
# We don't need to check out_of_order_processed here, because
1180
# the lookup_by_file_id will be removing anything processed
1181
# from the extras cache
1182
other_extra.pop(file_id)
1183
other_values = [(None, None) for i in xrange(idx)]
1184
other_values.append((other_path, other_ie))
1185
for alt_idx, alt_extra in enumerate(self._others_extra[idx+1:]):
1186
alt_idx = alt_idx + idx + 1
1187
alt_extra = self._others_extra[alt_idx]
1188
alt_tree = self._other_trees[alt_idx]
1189
other_values.append(self._lookup_by_file_id(
1190
alt_extra, alt_tree, file_id))
1191
yield other_path, file_id, None, other_values
added
removed
bzrlib/tree.py
