31
30
revision as _mod_revision,
35
33
from bzrlib.decorators import needs_read_lock
36
from bzrlib.errors import BzrError, BzrCheckError, NoSuchId
34
from bzrlib.errors import BzrError, NoSuchId
37
35
from bzrlib import errors
38
from bzrlib.inventory import Inventory, InventoryFile
36
from bzrlib.inventory import InventoryFile
39
37
from bzrlib.inter import InterObject
40
38
from bzrlib.osutils import fingerprint_file
41
39
import bzrlib.revision
42
40
from bzrlib.symbol_versioning import deprecated_function, deprecated_in
43
from bzrlib.trace import mutter, note
41
from bzrlib.trace import note
46
44
class Tree(object):
198
195
The yield order (ignoring root) would be::
199
196
a, f, a/b, a/d, a/b/c, a/d/e, f/g
198
:param yield_parents: If True, yield the parents from the root leading
199
down to specific_file_ids that have been requested. This has no
200
impact if specific_file_ids is None.
201
202
return self.inventory.iter_entries_by_dir(
202
specific_file_ids=specific_file_ids)
203
specific_file_ids=specific_file_ids, yield_parents=yield_parents)
204
205
def iter_references(self):
205
for path, entry in self.iter_entries_by_dir():
206
if entry.kind == 'tree-reference':
207
yield path, entry.file_id
206
if self.supports_tree_reference():
207
for path, entry in self.iter_entries_by_dir():
208
if entry.kind == 'tree-reference':
209
yield path, entry.file_id
209
211
def kind(self, file_id):
210
212
raise NotImplementedError("Tree subclass %s must implement kind"
221
223
def path_content_summary(self, path):
222
224
"""Get a summary of the information about path.
226
All the attributes returned are for the canonical form, not the
227
convenient form (if content filters are in use.)
224
229
:param path: A relative path within the tree.
225
230
:return: A tuple containing kind, size, exec, sha1-or-link.
226
231
Kind is always present (see tree.kind()).
227
size is present if kind is file, None otherwise.
232
size is present if kind is file and the size of the
233
canonical form can be cheaply determined, None otherwise.
228
234
exec is None unless kind is file and the platform supports the 'x'
230
236
sha1-or-link is the link target if kind is symlink, or the sha1 if
263
269
raise NotImplementedError(self.get_file)
271
def get_file_with_stat(self, file_id, path=None):
272
"""Get a file handle and stat object for file_id.
274
The default implementation returns (self.get_file, None) for backwards
277
:param file_id: The file id to read.
278
:param path: The path of the file, if it is known.
279
:return: A tuple (file_handle, stat_value_or_None). If the tree has
280
no stat facility, or need for a stat cache feedback during commit,
281
it may return None for the second element of the tuple.
283
return (self.get_file(file_id, path), None)
265
285
def get_file_text(self, file_id, path=None):
266
286
"""Return the byte content of a file.
385
405
bit_iter = iter(path.split("/"))
386
406
for elt in bit_iter:
387
407
lelt = elt.lower()
388
409
for child in self.iter_children(cur_id):
411
# XXX: it seem like if the child is known to be in the
412
# tree, we shouldn't need to go from its id back to
413
# its path -- mbp 2010-02-11
415
# XXX: it seems like we could be more efficient
416
# by just directly looking up the original name and
417
# only then searching all children; also by not
418
# chopping paths so much. -- mbp 2010-02-11
390
419
child_base = os.path.basename(self.id2path(child))
391
if child_base.lower() == lelt:
420
if (child_base == elt):
421
# if we found an exact match, we can stop now; if
422
# we found an approximate match we need to keep
423
# searching because there might be an exact match
393
cur_path = osutils.pathjoin(cur_path, child_base)
426
new_path = osutils.pathjoin(cur_path, child_base)
428
elif child_base.lower() == lelt:
430
new_path = osutils.pathjoin(cur_path, child_base)
396
432
# before a change is committed we can see this error...
399
437
# got to the end of this directory and no entries matched.
400
438
# Return what matched so far, plus the rest as specified.
831
870
will pass through to InterTree as appropriate.
873
# Formats that will be used to test this InterTree. If both are
874
# None, this InterTree will not be tested (e.g. because a complex
876
_matching_from_tree_format = None
877
_matching_to_tree_format = None
881
def _changes_from_entries(self, source_entry, target_entry,
882
source_path=None, target_path=None):
883
"""Generate a iter_changes tuple between source_entry and target_entry.
885
:param source_entry: An inventory entry from self.source, or None.
886
:param target_entry: An inventory entry from self.target, or None.
887
:param source_path: The path of source_entry, if known. If not known
888
it will be looked up.
889
:param target_path: The path of target_entry, if known. If not known
890
it will be looked up.
891
:return: A tuple, item 0 of which is an iter_changes result tuple, and
892
item 1 is True if there are any changes in the result tuple.
894
if source_entry is None:
895
if target_entry is None:
897
file_id = target_entry.file_id
899
file_id = source_entry.file_id
900
if source_entry is not None:
901
source_versioned = True
902
source_name = source_entry.name
903
source_parent = source_entry.parent_id
904
if source_path is None:
905
source_path = self.source.id2path(file_id)
906
source_kind, source_executable, source_stat = \
907
self.source._comparison_data(source_entry, source_path)
909
source_versioned = False
913
source_executable = None
914
if target_entry is not None:
915
target_versioned = True
916
target_name = target_entry.name
917
target_parent = target_entry.parent_id
918
if target_path is None:
919
target_path = self.target.id2path(file_id)
920
target_kind, target_executable, target_stat = \
921
self.target._comparison_data(target_entry, target_path)
923
target_versioned = False
927
target_executable = None
928
versioned = (source_versioned, target_versioned)
929
kind = (source_kind, target_kind)
930
changed_content = False
931
if source_kind != target_kind:
932
changed_content = True
933
elif source_kind == 'file':
934
if (self.source.get_file_sha1(file_id, source_path, source_stat) !=
935
self.target.get_file_sha1(file_id, target_path, target_stat)):
936
changed_content = True
937
elif source_kind == 'symlink':
938
if (self.source.get_symlink_target(file_id) !=
939
self.target.get_symlink_target(file_id)):
940
changed_content = True
941
# XXX: Yes, the indentation below is wrong. But fixing it broke
942
# test_merge.TestMergerEntriesLCAOnDisk.
943
# test_nested_tree_subtree_renamed_and_modified. We'll wait for
944
# the fix from bzr.dev -- vila 2009026
945
elif source_kind == 'tree-reference':
946
if (self.source.get_reference_revision(file_id, source_path)
947
!= self.target.get_reference_revision(file_id, target_path)):
948
changed_content = True
949
parent = (source_parent, target_parent)
950
name = (source_name, target_name)
951
executable = (source_executable, target_executable)
952
if (changed_content is not False or versioned[0] != versioned[1]
953
or parent[0] != parent[1] or name[0] != name[1] or
954
executable[0] != executable[1]):
958
return (file_id, (source_path, target_path), changed_content,
959
versioned, parent, name, kind, executable), changes
837
962
def compare(self, want_unchanged=False, specific_files=None,
838
963
extra_trees=None, require_versioned=False, include_root=False,
933
1074
# can be extras. So the fake_entry is solely used to look up
934
1075
# executable it values when execute is not supported.
935
1076
fake_entry = InventoryFile('unused', 'unused', 'unused')
936
for to_path, to_entry in to_entries_by_dir:
937
while all_unversioned and all_unversioned[0][0] < to_path.split('/'):
1077
for target_path, target_entry in to_entries_by_dir:
1078
while (all_unversioned and
1079
all_unversioned[0][0] < target_path.split('/')):
938
1080
unversioned_path = all_unversioned.popleft()
939
to_kind, to_executable, to_stat = \
1081
target_kind, target_executable, target_stat = \
940
1082
self.target._comparison_data(fake_entry, unversioned_path[1])
941
1083
yield (None, (None, unversioned_path[1]), True, (False, False),
943
1085
(None, unversioned_path[0][-1]),
945
(None, to_executable))
946
file_id = to_entry.file_id
947
to_paths[file_id] = to_path
1086
(None, target_kind),
1087
(None, target_executable))
1088
source_path, source_entry = from_data.get(target_entry.file_id,
1090
result, changes = self._changes_from_entries(source_entry,
1091
target_entry, source_path=source_path, target_path=target_path)
1092
to_paths[result[0]] = result[1][1]
948
1093
entry_count += 1
949
changed_content = False
950
from_path, from_entry = from_data.get(file_id, (None, None))
951
from_versioned = (from_entry is not None)
952
if from_entry is not None:
953
from_versioned = True
954
from_name = from_entry.name
955
from_parent = from_entry.parent_id
956
from_kind, from_executable, from_stat = \
957
self.source._comparison_data(from_entry, from_path)
958
1095
entry_count += 1
960
from_versioned = False
964
from_executable = None
965
versioned = (from_versioned, True)
966
to_kind, to_executable, to_stat = \
967
self.target._comparison_data(to_entry, to_path)
968
kind = (from_kind, to_kind)
969
if kind[0] != kind[1]:
970
changed_content = True
971
elif from_kind == 'file':
972
if (self.source.get_file_sha1(file_id, from_path, from_stat) !=
973
self.target.get_file_sha1(file_id, to_path, to_stat)):
974
changed_content = True
975
elif from_kind == 'symlink':
976
if (self.source.get_symlink_target(file_id) !=
977
self.target.get_symlink_target(file_id)):
978
changed_content = True
979
# XXX: Yes, the indentation below is wrong. But fixing it broke
980
# test_merge.TestMergerEntriesLCAOnDisk.
981
# test_nested_tree_subtree_renamed_and_modified. We'll wait for
982
# the fix from bzr.dev -- vila 2009026
983
elif from_kind == 'tree-reference':
984
if (self.source.get_reference_revision(file_id, from_path)
985
!= self.target.get_reference_revision(file_id, to_path)):
986
changed_content = True
987
parent = (from_parent, to_entry.parent_id)
988
name = (from_name, to_entry.name)
989
executable = (from_executable, to_executable)
990
1096
if pb is not None:
991
1097
pb.update('comparing files', entry_count, num_entries)
992
if (changed_content is not False or versioned[0] != versioned[1]
993
or parent[0] != parent[1] or name[0] != name[1] or
994
executable[0] != executable[1] or include_unchanged):
995
yield (file_id, (from_path, to_path), changed_content,
996
versioned, parent, name, kind, executable)
1098
if changes or include_unchanged:
1099
if specific_file_ids is not None:
1100
new_parent_id = result[4][1]
1101
precise_file_ids.add(new_parent_id)
1102
changed_file_ids.append(result[0])
1104
# Ensure correct behaviour for reparented/added specific files.
1105
if specific_files is not None:
1106
# Record output dirs
1107
if result[6][1] == 'directory':
1108
seen_dirs.add(result[0])
1109
# Record parents of reparented/added entries.
1110
versioned = result[3]
1112
if not versioned[0] or parents[0] != parents[1]:
1113
seen_parents.add(parents[1])
998
1114
while all_unversioned:
999
1115
# yield any trailing unversioned paths
1000
1116
unversioned_path = all_unversioned.popleft()
1005
1121
(None, unversioned_path[0][-1]),
1006
1122
(None, to_kind),
1007
1123
(None, to_executable))
1009
def get_to_path(to_entry):
1010
if to_entry.parent_id is None:
1011
to_path = '' # the root
1013
if to_entry.parent_id not in to_paths:
1015
return get_to_path(self.target.inventory[to_entry.parent_id])
1016
to_path = osutils.pathjoin(to_paths[to_entry.parent_id],
1018
to_paths[to_entry.file_id] = to_path
1124
# Yield all remaining source paths
1021
1125
for path, from_entry in from_entries_by_dir:
1022
1126
file_id = from_entry.file_id
1023
1127
if file_id in to_paths:
1024
1128
# already returned
1026
if not file_id in self.target.all_file_ids():
1130
if file_id not in self.target.all_file_ids():
1027
1131
# common case - paths we have not emitted are not present in
1031
to_path = get_to_path(self.target.inventory[file_id])
1135
to_path = self.target.id2path(file_id)
1032
1136
entry_count += 1
1033
1137
if pb is not None:
1034
1138
pb.update('comparing files', entry_count, num_entries)
1041
1145
executable = (from_executable, None)
1042
1146
changed_content = from_kind is not None
1043
1147
# the parent's path is necessarily known at this point.
1148
changed_file_ids.append(file_id)
1044
1149
yield(file_id, (path, to_path), changed_content, versioned, parent,
1045
1150
name, kind, executable)
1151
changed_file_ids = set(changed_file_ids)
1152
if specific_file_ids is not None:
1153
for result in self._handle_precise_ids(precise_file_ids,
1157
def _get_entry(self, tree, file_id):
1158
"""Get an inventory entry from a tree, with missing entries as None.
1160
If the tree raises NotImplementedError on accessing .inventory, then
1161
this is worked around using iter_entries_by_dir on just the file id
1164
:param tree: The tree to lookup the entry in.
1165
:param file_id: The file_id to lookup.
1168
inventory = tree.inventory
1169
except NotImplementedError:
1170
# No inventory available.
1172
iterator = tree.iter_entries_by_dir(specific_file_ids=[file_id])
1173
return iterator.next()[1]
1174
except StopIteration:
1178
return inventory[file_id]
1179
except errors.NoSuchId:
1182
def _handle_precise_ids(self, precise_file_ids, changed_file_ids,
1183
discarded_changes=None):
1184
"""Fill out a partial iter_changes to be consistent.
1186
:param precise_file_ids: The file ids of parents that were seen during
1188
:param changed_file_ids: The file ids of already emitted items.
1189
:param discarded_changes: An optional dict of precalculated
1190
iter_changes items which the partial iter_changes had not output
1192
:return: A generator of iter_changes items to output.
1194
# process parents of things that had changed under the users
1195
# requested paths to prevent incorrect paths or parent ids which
1196
# aren't in the tree.
1197
while precise_file_ids:
1198
precise_file_ids.discard(None)
1199
# Don't emit file_ids twice
1200
precise_file_ids.difference_update(changed_file_ids)
1201
if not precise_file_ids:
1203
# If the there was something at a given output path in source, we
1204
# have to include the entry from source in the delta, or we would
1205
# be putting this entry into a used path.
1207
for parent_id in precise_file_ids:
1209
paths.append(self.target.id2path(parent_id))
1210
except errors.NoSuchId:
1211
# This id has been dragged in from the source by delta
1212
# expansion and isn't present in target at all: we don't
1213
# need to check for path collisions on it.
1216
old_id = self.source.path2id(path)
1217
precise_file_ids.add(old_id)
1218
precise_file_ids.discard(None)
1219
current_ids = precise_file_ids
1220
precise_file_ids = set()
1221
# We have to emit all of precise_file_ids that have been altered.
1222
# We may have to output the children of some of those ids if any
1223
# directories have stopped being directories.
1224
for file_id in current_ids:
1226
if discarded_changes:
1227
result = discarded_changes.get(file_id)
1232
old_entry = self._get_entry(self.source, file_id)
1233
new_entry = self._get_entry(self.target, file_id)
1234
result, changes = self._changes_from_entries(
1235
old_entry, new_entry)
1238
# Get this parents parent to examine.
1239
new_parent_id = result[4][1]
1240
precise_file_ids.add(new_parent_id)
1242
if (result[6][0] == 'directory' and
1243
result[6][1] != 'directory'):
1244
# This stopped being a directory, the old children have
1246
if old_entry is None:
1247
# Reusing a discarded change.
1248
old_entry = self._get_entry(self.source, file_id)
1249
for child in old_entry.children.values():
1250
precise_file_ids.add(child.file_id)
1251
changed_file_ids.add(result[0])
1048
1255
class MultiWalker(object):