« back to all changes in this revision
Viewing changes to bzrlib/inventory.py
- browse files at revision 1914
- compare with another revision
- download diff
- download tarball
- view history from revision 1914
- Committer: Aaron Bentley
- Date: 2009-04-07 14:52:23 UTC
- mfrom: (1731.1744.1012 +trunk)
- Revision ID: aaron@aaronbentley.com-20090407145223-51piyz0wac8phofo
Merge bzr.dev into compossite-tree.
- files added:
- bzrlib/_chk_map_py.py
- bzrlib/tests/per_repository_chk
- files modified:
- .bzrignore
added
removed
bzrlib/inventory.py
38
38
39
39
import bzrlib
40
40
from bzrlib import (
41
chk_map,
41
42
errors,
42
43
generate_ids,
43
44
osutils,
90
91
>>> i.add(InventoryDirectory('123', 'src', ROOT_ID))
91
92
InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None)
92
93
>>> i.add(InventoryFile('2323', 'hello.c', parent_id='123'))
93
InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None)
94
InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None, revision=None)
94
95
>>> shouldbe = {0: '', 1: 'src', 2: 'src/hello.c'}
95
96
>>> for ix, j in enumerate(i.iter_entries()):
96
97
... print (j[0] == shouldbe[ix], j[1])
97
98
...
98
99
(True, InventoryDirectory('TREE_ROOT', u'', parent_id=None, revision=None))
99
100
(True, InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None))
100
(True, InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None))
101
(True, InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None, revision=None))
101
102
>>> i.add(InventoryFile('2324', 'bye.c', '123'))
102
InventoryFile('2324', 'bye.c', parent_id='123', sha1=None, len=None)
103
InventoryFile('2324', 'bye.c', parent_id='123', sha1=None, len=None, revision=None)
103
104
>>> i.add(InventoryDirectory('2325', 'wibble', '123'))
104
105
InventoryDirectory('2325', 'wibble', parent_id='123', revision=None)
105
106
>>> i.path2id('src/wibble')
107
108
>>> '2325' in i
108
109
True
109
110
>>> i.add(InventoryFile('2326', 'wibble.c', '2325'))
110
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
111
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None, revision=None)
111
112
>>> i['2326']
112
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
113
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None, revision=None)
113
114
>>> for path, entry in i.iter_entries():
114
115
... print path
115
116
...
566
567
self.executable = work_tree.is_executable(self.file_id, path=path)
567
568
568
569
def __repr__(self):
569
return ("%s(%r, %r, parent_id=%r, sha1=%r, len=%s)"
570
return ("%s(%r, %r, parent_id=%r, sha1=%r, len=%s, revision=%s)"
570
571
% (self.__class__.__name__,
571
572
self.file_id,
572
573
self.name,
573
574
self.parent_id,
574
575
self.text_sha1,
575
self.text_size))
576
self.text_size,
577
self.revision))
576
578
577
579
def _forget_tree_state(self):
578
580
self.text_sha1 = None
711
713
return compatible
712
714
713
715
714
class Inventory(object):
716
class CommonInventory(object):
717
"""Basic inventory logic, defined in terms of primitives like has_id."""
718
719
def __contains__(self, file_id):
720
"""True if this entry contains a file with given id.
721
722
>>> inv = Inventory()
723
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
724
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
725
>>> '123' in inv
726
True
727
>>> '456' in inv
728
False
729
730
Note that this method along with __iter__ are not encouraged for use as
731
they are less clear than specific query methods - they may be rmeoved
732
in the future.
733
"""
734
return self.has_id(file_id)
735
736
def id2path(self, file_id):
737
"""Return as a string the path to file_id.
738
739
>>> i = Inventory()
740
>>> e = i.add(InventoryDirectory('src-id', 'src', ROOT_ID))
741
>>> e = i.add(InventoryFile('foo-id', 'foo.c', parent_id='src-id'))
742
>>> print i.id2path('foo-id')
743
src/foo.c
744
"""
745
# get all names, skipping root
746
return '/'.join(reversed(
747
[parent.name for parent in
748
self._iter_file_id_parents(file_id)][:-1]))
749
750
def iter_entries(self, from_dir=None):
751
"""Return (path, entry) pairs, in order by name."""
752
if from_dir is None:
753
if self.root is None:
754
return
755
from_dir = self.root
756
yield '', self.root
757
elif isinstance(from_dir, basestring):
758
from_dir = self[from_dir]
759
760
# unrolling the recursive called changed the time from
761
# 440ms/663ms (inline/total) to 116ms/116ms
762
children = from_dir.children.items()
763
children.sort()
764
children = collections.deque(children)
765
stack = [(u'', children)]
766
while stack:
767
from_dir_relpath, children = stack[-1]
768
769
while children:
770
name, ie = children.popleft()
771
772
# we know that from_dir_relpath never ends in a slash
773
# and 'f' doesn't begin with one, we can do a string op, rather
774
# than the checks of pathjoin(), though this means that all paths
775
# start with a slash
776
path = from_dir_relpath + '/' + name
777
778
yield path[1:], ie
779
780
if ie.kind != 'directory':
781
continue
782
783
# But do this child first
784
new_children = ie.children.items()
785
new_children.sort()
786
new_children = collections.deque(new_children)
787
stack.append((path, new_children))
788
# Break out of inner loop, so that we start outer loop with child
789
break
790
else:
791
# if we finished all children, pop it off the stack
792
stack.pop()
793
794
def iter_entries_by_dir(self, from_dir=None, specific_file_ids=None,
795
yield_parents=False):
796
"""Iterate over the entries in a directory first order.
797
798
This returns all entries for a directory before returning
799
the entries for children of a directory. This is not
800
lexicographically sorted order, and is a hybrid between
801
depth-first and breadth-first.
802
803
:param yield_parents: If True, yield the parents from the root leading
804
down to specific_file_ids that have been requested. This has no
805
impact if specific_file_ids is None.
806
:return: This yields (path, entry) pairs
807
"""
808
if specific_file_ids and not isinstance(specific_file_ids, set):
809
specific_file_ids = set(specific_file_ids)
810
# TODO? Perhaps this should return the from_dir so that the root is
811
# yielded? or maybe an option?
812
if from_dir is None:
813
if self.root is None:
814
return
815
# Optimize a common case
816
if (not yield_parents and specific_file_ids is not None and
817
len(specific_file_ids) == 1):
818
file_id = list(specific_file_ids)[0]
819
if file_id in self:
820
yield self.id2path(file_id), self[file_id]
821
return
822
from_dir = self.root
823
if (specific_file_ids is None or yield_parents or
824
self.root.file_id in specific_file_ids):
825
yield u'', self.root
826
elif isinstance(from_dir, basestring):
827
from_dir = self[from_dir]
828
829
if specific_file_ids is not None:
830
# TODO: jam 20070302 This could really be done as a loop rather
831
# than a bunch of recursive calls.
832
parents = set()
833
byid = self
834
def add_ancestors(file_id):
835
if file_id not in byid:
836
return
837
parent_id = byid[file_id].parent_id
838
if parent_id is None:
839
return
840
if parent_id not in parents:
841
parents.add(parent_id)
842
add_ancestors(parent_id)
843
for file_id in specific_file_ids:
844
add_ancestors(file_id)
845
else:
846
parents = None
847
848
stack = [(u'', from_dir)]
849
while stack:
850
cur_relpath, cur_dir = stack.pop()
851
852
child_dirs = []
853
for child_name, child_ie in sorted(cur_dir.children.iteritems()):
854
855
child_relpath = cur_relpath + child_name
856
857
if (specific_file_ids is None or
858
child_ie.file_id in specific_file_ids or
859
(yield_parents and child_ie.file_id in parents)):
860
yield child_relpath, child_ie
861
862
if child_ie.kind == 'directory':
863
if parents is None or child_ie.file_id in parents:
864
child_dirs.append((child_relpath+'/', child_ie))
865
stack.extend(reversed(child_dirs))
866
867
def _make_delta(self, old):
868
"""Make an inventory delta from two inventories."""
869
old_ids = set(old)
870
new_ids = set(self)
871
adds = new_ids - old_ids
872
deletes = old_ids - new_ids
873
common = old_ids.intersection(new_ids)
874
delta = []
875
for file_id in deletes:
876
delta.append((old.id2path(file_id), None, file_id, None))
877
for file_id in adds:
878
delta.append((None, self.id2path(file_id), file_id, self[file_id]))
879
for file_id in common:
880
if old[file_id] != self[file_id]:
881
delta.append((old.id2path(file_id), self.id2path(file_id),
882
file_id, self[file_id]))
883
return delta
884
885
def _get_mutable_inventory(self):
886
"""Returns a mutable copy of the object.
887
888
Some inventories are immutable, yet working trees, for example, needs
889
to mutate exisiting inventories instead of creating a new one.
890
"""
891
raise NotImplementedError(self._get_mutable_inventory)
892
893
def make_entry(self, kind, name, parent_id, file_id=None):
894
"""Simple thunk to bzrlib.inventory.make_entry."""
895
return make_entry(kind, name, parent_id, file_id)
896
897
def entries(self):
898
"""Return list of (path, ie) for all entries except the root.
899
900
This may be faster than iter_entries.
901
"""
902
accum = []
903
def descend(dir_ie, dir_path):
904
kids = dir_ie.children.items()
905
kids.sort()
906
for name, ie in kids:
907
child_path = osutils.pathjoin(dir_path, name)
908
accum.append((child_path, ie))
909
if ie.kind == 'directory':
910
descend(ie, child_path)
911
912
descend(self.root, u'')
913
return accum
914
915
def directories(self):
916
"""Return (path, entry) pairs for all directories, including the root.
917
"""
918
accum = []
919
def descend(parent_ie, parent_path):
920
accum.append((parent_path, parent_ie))
921
922
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
923
kids.sort()
924
925
for name, child_ie in kids:
926
child_path = osutils.pathjoin(parent_path, name)
927
descend(child_ie, child_path)
928
descend(self.root, u'')
929
return accum
930
931
def path2id(self, name):
932
"""Walk down through directories to return entry of last component.
933
934
names may be either a list of path components, or a single
935
string, in which case it is automatically split.
936
937
This returns the entry of the last component in the path,
938
which may be either a file or a directory.
939
940
Returns None IFF the path is not found.
941
"""
942
if isinstance(name, basestring):
943
name = osutils.splitpath(name)
944
945
# mutter("lookup path %r" % name)
946
947
try:
948
parent = self.root
949
except errors.NoSuchId:
950
# root doesn't exist yet so nothing else can
951
return None
952
if parent is None:
953
return None
954
for f in name:
955
try:
956
children = getattr(parent, 'children', None)
957
if children is None:
958
return None
959
cie = children[f]
960
parent = cie
961
except KeyError:
962
# or raise an error?
963
return None
964
965
return parent.file_id
966
967
def filter(self, specific_fileids):
968
"""Get an inventory view filtered against a set of file-ids.
969
970
Children of directories and parents are included.
971
972
The result may or may not reference the underlying inventory
973
so it should be treated as immutable.
974
"""
975
interesting_parents = set()
976
for fileid in specific_fileids:
977
try:
978
interesting_parents.update(self.get_idpath(fileid))
979
except errors.NoSuchId:
980
# This fileid is not in the inventory - that's ok
981
pass
982
entries = self.iter_entries()
983
if self.root is None:
984
return Inventory(root_id=None)
985
other = Inventory(entries.next()[1].file_id)
986
other.root.revision = self.root.revision
987
other.revision_id = self.revision_id
988
directories_to_expand = set()
989
for path, entry in entries:
990
file_id = entry.file_id
991
if (file_id in specific_fileids
992
or entry.parent_id in directories_to_expand):
993
if entry.kind == 'directory':
994
directories_to_expand.add(file_id)
995
elif file_id not in interesting_parents:
996
continue
997
other.add(entry.copy())
998
return other
999
1000
def get_idpath(self, file_id):
1001
"""Return a list of file_ids for the path to an entry.
1002
1003
The list contains one element for each directory followed by
1004
the id of the file itself. So the length of the returned list
1005
is equal to the depth of the file in the tree, counting the
1006
root directory as depth 1.
1007
"""
1008
p = []
1009
for parent in self._iter_file_id_parents(file_id):
1010
p.insert(0, parent.file_id)
1011
return p
1012
1013
1014
class Inventory(CommonInventory):
715
1015
"""Inventory of versioned files in a tree.
716
1016
717
1017
This describes which file_id is present at each point in the tree,
730
1030
731
1031
>>> inv = Inventory()
732
1032
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
733
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
1033
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
734
1034
>>> inv['123-123'].name
735
1035
'hello.c'
736
1036
750
1050
Traceback (most recent call last):
751
1051
BzrError: parent_id {TREE_ROOT} not in inventory
752
1052
>>> inv.add(InventoryFile('123-123', 'hello.c', 'TREE_ROOT-12345678-12345678'))
753
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT-12345678-12345678', sha1=None, len=None)
1053
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT-12345678-12345678', sha1=None, len=None, revision=None)
754
1054
"""
755
1055
def __init__(self, root_id=ROOT_ID, revision_id=None):
756
1056
"""Create or read an inventory.
852
1152
# Pop the child which to allow detection of children whose
853
1153
# parents were deleted and which were not reattached to a new
854
1154
# parent.
855
new_entry.children = children.pop(new_entry.file_id, {})
1155
replacement = InventoryDirectory(new_entry.file_id,
1156
new_entry.name, new_entry.parent_id)
1157
replacement.revision = new_entry.revision
1158
replacement.children = children.pop(replacement.file_id, {})
1159
new_entry = replacement
856
1160
self.add(new_entry)
857
1161
if len(children):
858
1162
# Get the parent id that was deleted
878
1182
return other
879
1183
880
1184
def _get_mutable_inventory(self):
881
"""Returns a mutable copy of the object.
882
883
Some inventories are immutable, yet working trees, for example, needs
884
to mutate exisiting inventories instead of creating a new one.
885
"""
1185
"""See CommonInventory._get_mutable_inventory."""
886
1186
return deepcopy(self)
887
1187
888
1188
def __iter__(self):
1189
"""Iterate over all file-ids."""
889
1190
return iter(self._byid)
890
1191
1192
def iter_just_entries(self):
1193
"""Iterate over all entries.
1194
1195
Unlike iter_entries(), just the entries are returned (not (path, ie))
1196
and the order of entries is undefined.
1197
1198
XXX: We may not want to merge this into bzr.dev.
1199
"""
1200
if self.root is None:
1201
return
1202
for _, ie in self._byid.iteritems():
1203
yield ie
1204
891
1205
def __len__(self):
892
1206
"""Returns number of entries."""
893
1207
return len(self._byid)
894
1208
895
def iter_entries(self, from_dir=None):
896
"""Return (path, entry) pairs, in order by name."""
897
if from_dir is None:
898
if self.root is None:
899
return
900
from_dir = self.root
901
yield '', self.root
902
elif isinstance(from_dir, basestring):
903
from_dir = self._byid[from_dir]
904
905
# unrolling the recursive called changed the time from
906
# 440ms/663ms (inline/total) to 116ms/116ms
907
children = from_dir.children.items()
908
children.sort()
909
children = collections.deque(children)
910
stack = [(u'', children)]
911
while stack:
912
from_dir_relpath, children = stack[-1]
913
914
while children:
915
name, ie = children.popleft()
916
917
# we know that from_dir_relpath never ends in a slash
918
# and 'f' doesn't begin with one, we can do a string op, rather
919
# than the checks of pathjoin(), though this means that all paths
920
# start with a slash
921
path = from_dir_relpath + '/' + name
922
923
yield path[1:], ie
924
925
if ie.kind != 'directory':
926
continue
927
928
# But do this child first
929
new_children = ie.children.items()
930
new_children.sort()
931
new_children = collections.deque(new_children)
932
stack.append((path, new_children))
933
# Break out of inner loop, so that we start outer loop with child
934
break
935
else:
936
# if we finished all children, pop it off the stack
937
stack.pop()
938
939
def iter_entries_by_dir(self, from_dir=None, specific_file_ids=None,
940
yield_parents=False):
941
"""Iterate over the entries in a directory first order.
942
943
This returns all entries for a directory before returning
944
the entries for children of a directory. This is not
945
lexicographically sorted order, and is a hybrid between
946
depth-first and breadth-first.
947
948
:param yield_parents: If True, yield the parents from the root leading
949
down to specific_file_ids that have been requested. This has no
950
impact if specific_file_ids is None.
951
:return: This yields (path, entry) pairs
952
"""
953
if specific_file_ids and not isinstance(specific_file_ids, set):
954
specific_file_ids = set(specific_file_ids)
955
# TODO? Perhaps this should return the from_dir so that the root is
956
# yielded? or maybe an option?
957
if from_dir is None:
958
if self.root is None:
959
return
960
# Optimize a common case
961
if (not yield_parents and specific_file_ids is not None and
962
len(specific_file_ids) == 1):
963
file_id = list(specific_file_ids)[0]
964
if file_id in self:
965
yield self.id2path(file_id), self[file_id]
966
return
967
from_dir = self.root
968
if (specific_file_ids is None or yield_parents or
969
self.root.file_id in specific_file_ids):
970
yield u'', self.root
971
elif isinstance(from_dir, basestring):
972
from_dir = self._byid[from_dir]
973
974
if specific_file_ids is not None:
975
# TODO: jam 20070302 This could really be done as a loop rather
976
# than a bunch of recursive calls.
977
parents = set()
978
byid = self._byid
979
def add_ancestors(file_id):
980
if file_id not in byid:
981
return
982
parent_id = byid[file_id].parent_id
983
if parent_id is None:
984
return
985
if parent_id not in parents:
986
parents.add(parent_id)
987
add_ancestors(parent_id)
988
for file_id in specific_file_ids:
989
add_ancestors(file_id)
990
else:
991
parents = None
992
993
stack = [(u'', from_dir)]
994
while stack:
995
cur_relpath, cur_dir = stack.pop()
996
997
child_dirs = []
998
for child_name, child_ie in sorted(cur_dir.children.iteritems()):
999
1000
child_relpath = cur_relpath + child_name
1001
1002
if (specific_file_ids is None or
1003
child_ie.file_id in specific_file_ids or
1004
(yield_parents and child_ie.file_id in parents)):
1005
yield child_relpath, child_ie
1006
1007
if child_ie.kind == 'directory':
1008
if parents is None or child_ie.file_id in parents:
1009
child_dirs.append((child_relpath+'/', child_ie))
1010
stack.extend(reversed(child_dirs))
1011
1012
def make_entry(self, kind, name, parent_id, file_id=None):
1013
"""Simple thunk to bzrlib.inventory.make_entry."""
1014
return make_entry(kind, name, parent_id, file_id)
1015
1016
def entries(self):
1017
"""Return list of (path, ie) for all entries except the root.
1018
1019
This may be faster than iter_entries.
1020
"""
1021
accum = []
1022
def descend(dir_ie, dir_path):
1023
kids = dir_ie.children.items()
1024
kids.sort()
1025
for name, ie in kids:
1026
child_path = osutils.pathjoin(dir_path, name)
1027
accum.append((child_path, ie))
1028
if ie.kind == 'directory':
1029
descend(ie, child_path)
1030
1031
descend(self.root, u'')
1032
return accum
1033
1034
def directories(self):
1035
"""Return (path, entry) pairs for all directories, including the root.
1036
"""
1037
accum = []
1038
def descend(parent_ie, parent_path):
1039
accum.append((parent_path, parent_ie))
1040
1041
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
1042
kids.sort()
1043
1044
for name, child_ie in kids:
1045
child_path = osutils.pathjoin(parent_path, name)
1046
descend(child_ie, child_path)
1047
descend(self.root, u'')
1048
return accum
1049
1050
def __contains__(self, file_id):
1051
"""True if this entry contains a file with given id.
1052
1053
>>> inv = Inventory()
1054
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1055
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
1056
>>> '123' in inv
1057
True
1058
>>> '456' in inv
1059
False
1060
"""
1061
return (file_id in self._byid)
1062
1063
1209
def __getitem__(self, file_id):
1064
1210
"""Return the entry for given file_id.
1065
1211
1066
1212
>>> inv = Inventory()
1067
1213
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
1068
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
1214
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
1069
1215
>>> inv['123123'].name
1070
1216
'hello.c'
1071
1217
"""
1147
1293
1148
1294
>>> inv = Inventory()
1149
1295
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1150
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
1296
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
1151
1297
>>> '123' in inv
1152
1298
True
1153
1299
>>> del inv['123']
1167
1313
>>> i1 == i2
1168
1314
True
1169
1315
>>> i1.add(InventoryFile('123', 'foo', ROOT_ID))
1170
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
1316
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
1171
1317
>>> i1 == i2
1172
1318
False
1173
1319
>>> i2.add(InventoryFile('123', 'foo', ROOT_ID))
1174
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
1320
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
1175
1321
>>> i1 == i2
1176
1322
True
1177
1323
"""
1196
1342
yield ie
1197
1343
file_id = ie.parent_id
1198
1344
1199
def get_idpath(self, file_id):
1200
"""Return a list of file_ids for the path to an entry.
1201
1202
The list contains one element for each directory followed by
1203
the id of the file itself. So the length of the returned list
1204
is equal to the depth of the file in the tree, counting the
1205
root directory as depth 1.
1206
"""
1207
p = []
1208
for parent in self._iter_file_id_parents(file_id):
1209
p.insert(0, parent.file_id)
1210
return p
1211
1212
def id2path(self, file_id):
1213
"""Return as a string the path to file_id.
1214
1215
>>> i = Inventory()
1216
>>> e = i.add(InventoryDirectory('src-id', 'src', ROOT_ID))
1217
>>> e = i.add(InventoryFile('foo-id', 'foo.c', parent_id='src-id'))
1218
>>> print i.id2path('foo-id')
1219
src/foo.c
1220
"""
1221
# get all names, skipping root
1222
return '/'.join(reversed(
1223
[parent.name for parent in
1224
self._iter_file_id_parents(file_id)][:-1]))
1225
1226
def path2id(self, name):
1227
"""Walk down through directories to return entry of last component.
1228
1229
names may be either a list of path components, or a single
1230
string, in which case it is automatically split.
1231
1232
This returns the entry of the last component in the path,
1233
which may be either a file or a directory.
1234
1235
Returns None IFF the path is not found.
1236
"""
1237
if isinstance(name, basestring):
1238
name = osutils.splitpath(name)
1239
1240
# mutter("lookup path %r" % name)
1241
1242
parent = self.root
1243
if parent is None:
1244
return None
1245
for f in name:
1246
try:
1247
children = getattr(parent, 'children', None)
1248
if children is None:
1249
return None
1250
cie = children[f]
1251
parent = cie
1252
except KeyError:
1253
# or raise an error?
1254
return None
1255
1256
return parent.file_id
1257
1258
1345
def has_filename(self, names):
1259
1346
return bool(self.path2id(names))
1260
1347
1346
1433
def is_root(self, file_id):
1347
1434
return self.root is not None and file_id == self.root.file_id
1348
1435
1349
def filter(self, specific_fileids):
1350
"""Get an inventory view filtered against a set of file-ids.
1351
1352
Children of directories and parents are included.
1353
1354
The result may or may not reference the underlying inventory
1355
so it should be treated as immutable.
1356
"""
1357
interesting_parents = set()
1358
for fileid in specific_fileids:
1436
1437
class CHKInventory(CommonInventory):
1438
"""An inventory persisted in a CHK store.
1439
1440
By design, a CHKInventory is immutable so many of the methods
1441
supported by Inventory - add, rename, apply_delta, etc - are *not*
1442
supported. To create a new CHKInventory, use create_by_apply_delta()
1443
or from_inventory(), say.
1444
1445
Internally, a CHKInventory has one or two CHKMaps:
1446
1447
* id_to_entry - a map from (file_id,) => InventoryEntry as bytes
1448
* parent_id_basename_to_file_id - a map from (parent_id, basename_utf8)
1449
=> file_id as bytes
1450
1451
The second map is optional and not present in early CHkRepository's.
1452
1453
No caching is performed: every method call or item access will perform
1454
requests to the storage layer. As such, keep references to objects you
1455
want to reuse.
1456
"""
1457
1458
def __init__(self, search_key_name):
1459
CommonInventory.__init__(self)
1460
self._fileid_to_entry_cache = {}
1461
self._path_to_fileid_cache = {}
1462
self._search_key_name = search_key_name
1463
1464
def _entry_to_bytes(self, entry):
1465
"""Serialise entry as a single bytestring.
1466
1467
:param Entry: An inventory entry.
1468
:return: A bytestring for the entry.
1469
1470
The BNF:
1471
ENTRY ::= FILE | DIR | SYMLINK | TREE
1472
FILE ::= "file: " COMMON SEP SHA SEP SIZE SEP EXECUTABLE
1473
DIR ::= "dir: " COMMON
1474
SYMLINK ::= "symlink: " COMMON SEP TARGET_UTF8
1475
TREE ::= "tree: " COMMON REFERENCE_REVISION
1476
COMMON ::= FILE_ID SEP PARENT_ID SEP NAME_UTF8 SEP REVISION
1477
SEP ::= "\n"
1478
"""
1479
if entry.parent_id is not None:
1480
parent_str = entry.parent_id
1481
else:
1482
parent_str = ''
1483
name_str = entry.name.encode("utf8")
1484
if entry.kind == 'file':
1485
if entry.executable:
1486
exec_str = "Y"
1487
else:
1488
exec_str = "N"
1489
return "file: %s\n%s\n%s\n%s\n%s\n%d\n%s" % (
1490
entry.file_id, parent_str, name_str, entry.revision,
1491
entry.text_sha1, entry.text_size, exec_str)
1492
elif entry.kind == 'directory':
1493
return "dir: %s\n%s\n%s\n%s" % (
1494
entry.file_id, parent_str, name_str, entry.revision)
1495
elif entry.kind == 'symlink':
1496
return "symlink: %s\n%s\n%s\n%s\n%s" % (
1497
entry.file_id, parent_str, name_str, entry.revision,
1498
entry.symlink_target.encode("utf8"))
1499
elif entry.kind == 'tree-reference':
1500
return "tree: %s\n%s\n%s\n%s\n%s" % (
1501
entry.file_id, parent_str, name_str, entry.revision,
1502
entry.reference_revision)
1503
else:
1504
raise ValueError("unknown kind %r" % entry.kind)
1505
1506
@staticmethod
1507
def _bytes_to_utf8name_key(bytes):
1508
"""Get the file_id, revision_id key out of bytes."""
1509
# We don't normally care about name, except for times when we want
1510
# to filter out empty names because of non rich-root...
1511
sections = bytes.split('\n')
1512
kind, file_id = sections[0].split(': ')
1513
return (sections[2], file_id, sections[3])
1514
1515
def _bytes_to_entry(self, bytes):
1516
"""Deserialise a serialised entry."""
1517
sections = bytes.split('\n')
1518
if sections[0].startswith("file: "):
1519
result = InventoryFile(sections[0][6:],
1520
sections[2].decode('utf8'),
1521
sections[1])
1522
result.text_sha1 = sections[4]
1523
result.text_size = int(sections[5])
1524
result.executable = sections[6] == "Y"
1525
elif sections[0].startswith("dir: "):
1526
result = CHKInventoryDirectory(sections[0][5:],
1527
sections[2].decode('utf8'),
1528
sections[1], self)
1529
elif sections[0].startswith("symlink: "):
1530
result = InventoryLink(sections[0][9:],
1531
sections[2].decode('utf8'),
1532
sections[1])
1533
result.symlink_target = sections[4].decode('utf8')
1534
elif sections[0].startswith("tree: "):
1535
result = TreeReference(sections[0][6:],
1536
sections[2].decode('utf8'),
1537
sections[1])
1538
result.reference_revision = sections[4]
1539
else:
1540
raise ValueError("Not a serialised entry %r" % bytes)
1541
result.revision = sections[3]
1542
if result.parent_id == '':
1543
result.parent_id = None
1544
self._fileid_to_entry_cache[result.file_id] = result
1545
return result
1546
1547
def _get_mutable_inventory(self):
1548
"""See CommonInventory._get_mutable_inventory."""
1549
entries = self.iter_entries()
1550
if self.root_id is not None:
1551
entries.next()
1552
inv = Inventory(self.root_id, self.revision_id)
1553
for path, inv_entry in entries:
1554
inv.add(inv_entry)
1555
return inv
1556
1557
def create_by_apply_delta(self, inventory_delta, new_revision_id,
1558
propagate_caches=False):
1559
"""Create a new CHKInventory by applying inventory_delta to this one.
1560
1561
:param inventory_delta: The inventory delta to apply. See
1562
Inventory.apply_delta for details.
1563
:param new_revision_id: The revision id of the resulting CHKInventory.
1564
:param propagate_caches: If True, the caches for this inventory are
1565
copied to and updated for the result.
1566
:return: The new CHKInventory.
1567
"""
1568
result = CHKInventory(self._search_key_name)
1569
if propagate_caches:
1570
# Just propagate the path-to-fileid cache for now
1571
result._path_to_fileid_cache = dict(self._path_to_fileid_cache.iteritems())
1572
search_key_func = chk_map.search_key_registry.get(self._search_key_name)
1573
self.id_to_entry._ensure_root()
1574
maximum_size = self.id_to_entry._root_node.maximum_size
1575
result.revision_id = new_revision_id
1576
result.id_to_entry = chk_map.CHKMap(
1577
self.id_to_entry._store,
1578
self.id_to_entry.key(),
1579
search_key_func=search_key_func)
1580
result.id_to_entry._ensure_root()
1581
result.id_to_entry._root_node.set_maximum_size(maximum_size)
1582
parent_id_basename_delta = []
1583
if self.parent_id_basename_to_file_id is not None:
1584
result.parent_id_basename_to_file_id = chk_map.CHKMap(
1585
self.parent_id_basename_to_file_id._store,
1586
self.parent_id_basename_to_file_id.key(),
1587
search_key_func=search_key_func)
1588
result.parent_id_basename_to_file_id._ensure_root()
1589
self.parent_id_basename_to_file_id._ensure_root()
1590
result_p_id_root = result.parent_id_basename_to_file_id._root_node
1591
p_id_root = self.parent_id_basename_to_file_id._root_node
1592
result_p_id_root.set_maximum_size(p_id_root.maximum_size)
1593
result_p_id_root._key_width = p_id_root._key_width
1594
else:
1595
result.parent_id_basename_to_file_id = None
1596
result.root_id = self.root_id
1597
id_to_entry_delta = []
1598
for old_path, new_path, file_id, entry in inventory_delta:
1599
# file id changes
1600
if new_path == '':
1601
result.root_id = file_id
1602
if new_path is None:
1603
# Make a delete:
1604
new_key = None
1605
new_value = None
1606
# Update caches
1607
if propagate_caches:
1608
try:
1609
del result._path_to_fileid_cache[old_path]
1610
except KeyError:
1611
pass
1612
else:
1613
new_key = (file_id,)
1614
new_value = result._entry_to_bytes(entry)
1615
# Update caches. It's worth doing this whether
1616
# we're propagating the old caches or not.
1617
result._path_to_fileid_cache[new_path] = file_id
1618
if old_path is None:
1619
old_key = None
1620
else:
1621
old_key = (file_id,)
1622
id_to_entry_delta.append((old_key, new_key, new_value))
1623
if result.parent_id_basename_to_file_id is not None:
1624
# parent_id, basename changes
1625
if old_path is None:
1626
old_key = None
1627
else:
1628
old_entry = self[file_id]
1629
old_key = self._parent_id_basename_key(old_entry)
1630
if new_path is None:
1631
new_key = None
1632
new_value = None
1633
else:
1634
new_key = self._parent_id_basename_key(entry)
1635
new_value = file_id
1636
if old_key != new_key:
1637
# If the two keys are the same, the value will be unchanged
1638
# as its always the file id.
1639
parent_id_basename_delta.append((old_key, new_key, new_value))
1640
result.id_to_entry.apply_delta(id_to_entry_delta)
1641
if parent_id_basename_delta:
1642
result.parent_id_basename_to_file_id.apply_delta(parent_id_basename_delta)
1643
return result
1644
1645
@classmethod
1646
def deserialise(klass, chk_store, bytes, expected_revision_id):
1647
"""Deserialise a CHKInventory.
1648
1649
:param chk_store: A CHK capable VersionedFiles instance.
1650
:param bytes: The serialised bytes.
1651
:param expected_revision_id: The revision ID we think this inventory is
1652
for.
1653
:return: A CHKInventory
1654
"""
1655
lines = bytes.split('\n')
1656
if lines[-1] != '':
1657
raise AssertionError('bytes to deserialize must end with an eol')
1658
lines.pop()
1659
if lines[0] != 'chkinventory:':
1660
raise ValueError("not a serialised CHKInventory: %r" % bytes)
1661
info = {}
1662
allowed_keys = frozenset(['root_id', 'revision_id', 'search_key_name',
1663
'parent_id_basename_to_file_id',
1664
'id_to_entry'])
1665
for line in lines[1:]:
1666
key, value = line.split(': ', 1)
1667
if key not in allowed_keys:
1668
raise errors.BzrError('Unknown key in inventory: %r\n%r'
1669
% (key, bytes))
1670
if key in info:
1671
raise errors.BzrError('Duplicate key in inventory: %r\n%r'
1672
% (key, bytes))
1673
info[key] = value
1674
revision_id = info['revision_id']
1675
root_id = info['root_id']
1676
search_key_name = info.get('search_key_name', 'plain')
1677
parent_id_basename_to_file_id = info.get(
1678
'parent_id_basename_to_file_id', None)
1679
id_to_entry = info['id_to_entry']
1680
1681
result = CHKInventory(search_key_name)
1682
result.revision_id = revision_id
1683
result.root_id = root_id
1684
search_key_func = chk_map.search_key_registry.get(
1685
result._search_key_name)
1686
if parent_id_basename_to_file_id is not None:
1687
result.parent_id_basename_to_file_id = chk_map.CHKMap(
1688
chk_store, (parent_id_basename_to_file_id,),
1689
search_key_func=search_key_func)
1690
else:
1691
result.parent_id_basename_to_file_id = None
1692
1693
result.id_to_entry = chk_map.CHKMap(chk_store, (id_to_entry,),
1694
search_key_func=search_key_func)
1695
if (result.revision_id,) != expected_revision_id:
1696
raise ValueError("Mismatched revision id and expected: %r, %r" %
1697
(result.revision_id, expected_revision_id))
1698
return result
1699
1700
@classmethod
1701
def from_inventory(klass, chk_store, inventory, maximum_size=0, search_key_name='plain'):
1702
"""Create a CHKInventory from an existing inventory.
1703
1704
The content of inventory is copied into the chk_store, and a
1705
CHKInventory referencing that is returned.
1706
1707
:param chk_store: A CHK capable VersionedFiles instance.
1708
:param inventory: The inventory to copy.
1709
:param maximum_size: The CHKMap node size limit.
1710
:param search_key_name: The identifier for the search key function
1711
"""
1712
result = CHKInventory(search_key_name)
1713
result.revision_id = inventory.revision_id
1714
result.root_id = inventory.root.file_id
1715
search_key_func = chk_map.search_key_registry.get(search_key_name)
1716
result.id_to_entry = chk_map.CHKMap(chk_store, None, search_key_func)
1717
result.id_to_entry._root_node.set_maximum_size(maximum_size)
1718
file_id_delta = []
1719
result.parent_id_basename_to_file_id = chk_map.CHKMap(chk_store,
1720
None, search_key_func)
1721
result.parent_id_basename_to_file_id._root_node.set_maximum_size(
1722
maximum_size)
1723
result.parent_id_basename_to_file_id._root_node._key_width = 2
1724
parent_id_delta = []
1725
for path, entry in inventory.iter_entries():
1726
file_id_delta.append((None, (entry.file_id,),
1727
result._entry_to_bytes(entry)))
1728
parent_id_delta.append(
1729
(None, result._parent_id_basename_key(entry),
1730
entry.file_id))
1731
result.id_to_entry.apply_delta(file_id_delta)
1732
result.parent_id_basename_to_file_id.apply_delta(parent_id_delta)
1733
return result
1734
1735
def _parent_id_basename_key(self, entry):
1736
"""Create a key for a entry in a parent_id_basename_to_file_id index."""
1737
if entry.parent_id is not None:
1738
parent_id = entry.parent_id
1739
else:
1740
parent_id = ''
1741
return parent_id, entry.name.encode('utf8')
1742
1743
def __getitem__(self, file_id):
1744
"""map a single file_id -> InventoryEntry."""
1745
if file_id is None:
1746
raise errors.NoSuchId(self, file_id)
1747
result = self._fileid_to_entry_cache.get(file_id, None)
1748
if result is not None:
1749
return result
1750
try:
1751
return self._bytes_to_entry(
1752
self.id_to_entry.iteritems([(file_id,)]).next()[1])
1753
except StopIteration:
1754
# really we're passing an inventory, not a tree...
1755
raise errors.NoSuchId(self, file_id)
1756
1757
def has_id(self, file_id):
1758
# Perhaps have an explicit 'contains' method on CHKMap ?
1759
if self._fileid_to_entry_cache.get(file_id, None) is not None:
1760
return True
1761
return len(list(self.id_to_entry.iteritems([(file_id,)]))) == 1
1762
1763
def is_root(self, file_id):
1764
return file_id == self.root_id
1765
1766
def _iter_file_id_parents(self, file_id):
1767
"""Yield the parents of file_id up to the root."""
1768
while file_id is not None:
1359
1769
try:
1360
interesting_parents.update(self.get_idpath(fileid))
1361
except errors.NoSuchId:
1362
# This fileid is not in the inventory - that's ok
1363
pass
1364
entries = self.iter_entries()
1365
if self.root is None:
1366
return Inventory(root_id=None)
1367
other = Inventory(entries.next()[1].file_id)
1368
other.root.revision = self.root.revision
1369
other.revision_id = self.revision_id
1370
directories_to_expand = set()
1371
for path, entry in entries:
1372
file_id = entry.file_id
1373
if (file_id in specific_fileids
1374
or entry.parent_id in directories_to_expand):
1375
if entry.kind == 'directory':
1376
directories_to_expand.add(file_id)
1377
elif file_id not in interesting_parents:
1770
ie = self[file_id]
1771
except KeyError:
1772
raise errors.NoSuchId(tree=self, file_id=file_id)
1773
yield ie
1774
file_id = ie.parent_id
1775
1776
def __iter__(self):
1777
"""Iterate over all file-ids."""
1778
for key, _ in self.id_to_entry.iteritems():
1779
yield key[-1]
1780
1781
def iter_just_entries(self):
1782
"""Iterate over all entries.
1783
1784
Unlike iter_entries(), just the entries are returned (not (path, ie))
1785
and the order of entries is undefined.
1786
1787
XXX: We may not want to merge this into bzr.dev.
1788
"""
1789
for key, entry in self.id_to_entry.iteritems():
1790
file_id = key[0]
1791
ie = self._fileid_to_entry_cache.get(file_id, None)
1792
if ie is None:
1793
ie = self._bytes_to_entry(entry)
1794
self._fileid_to_entry_cache[file_id] = ie
1795
yield ie
1796
1797
def iter_changes(self, basis):
1798
"""Generate a Tree.iter_changes change list between this and basis.
1799
1800
:param basis: Another CHKInventory.
1801
:return: An iterator over the changes between self and basis, as per
1802
tree.iter_changes().
1803
"""
1804
# We want: (file_id, (path_in_source, path_in_target),
1805
# changed_content, versioned, parent, name, kind,
1806
# executable)
1807
for key, basis_value, self_value in \
1808
self.id_to_entry.iter_changes(basis.id_to_entry):
1809
file_id = key[0]
1810
if basis_value is not None:
1811
basis_entry = basis._bytes_to_entry(basis_value)
1812
path_in_source = basis.id2path(file_id)
1813
basis_parent = basis_entry.parent_id
1814
basis_name = basis_entry.name
1815
basis_executable = basis_entry.executable
1816
else:
1817
path_in_source = None
1818
basis_parent = None
1819
basis_name = None
1820
basis_executable = None
1821
if self_value is not None:
1822
self_entry = self._bytes_to_entry(self_value)
1823
path_in_target = self.id2path(file_id)
1824
self_parent = self_entry.parent_id
1825
self_name = self_entry.name
1826
self_executable = self_entry.executable
1827
else:
1828
path_in_target = None
1829
self_parent = None
1830
self_name = None
1831
self_executable = None
1832
if basis_value is None:
1833
# add
1834
kind = (None, self_entry.kind)
1835
versioned = (False, True)
1836
elif self_value is None:
1837
# delete
1838
kind = (basis_entry.kind, None)
1839
versioned = (True, False)
1840
else:
1841
kind = (basis_entry.kind, self_entry.kind)
1842
versioned = (True, True)
1843
changed_content = False
1844
if kind[0] != kind[1]:
1845
changed_content = True
1846
elif kind[0] == 'file':
1847
if (self_entry.text_size != basis_entry.text_size or
1848
self_entry.text_sha1 != basis_entry.text_sha1):
1849
changed_content = True
1850
elif kind[0] == 'symlink':
1851
if self_entry.symlink_target != basis_entry.symlink_target:
1852
changed_content = True
1853
elif kind[0] == 'tree-reference':
1854
if (self_entry.reference_revision !=
1855
basis_entry.reference_revision):
1856
changed_content = True
1857
parent = (basis_parent, self_parent)
1858
name = (basis_name, self_name)
1859
executable = (basis_executable, self_executable)
1860
if (not changed_content
1861
and parent[0] == parent[1]
1862
and name[0] == name[1]
1863
and executable[0] == executable[1]):
1864
# Could happen when only the revision changed for a directory
1865
# for instance.
1378
1866
continue
1379
other.add(entry.copy())
1380
return other
1381
1867
yield (file_id, (path_in_source, path_in_target), changed_content,
1868
versioned, parent, name, kind, executable)
1869
1870
def __len__(self):
1871
"""Return the number of entries in the inventory."""
1872
return len(self.id_to_entry)
1873
1874
def _make_delta(self, old):
1875
"""Make an inventory delta from two inventories."""
1876
if type(old) != CHKInventory:
1877
return CommonInventory._make_delta(self, old)
1878
delta = []
1879
for key, old_value, self_value in \
1880
self.id_to_entry.iter_changes(old.id_to_entry):
1881
file_id = key[0]
1882
if old_value is not None:
1883
old_path = old.id2path(file_id)
1884
else:
1885
old_path = None
1886
if self_value is not None:
1887
entry = self._bytes_to_entry(self_value)
1888
self._fileid_to_entry_cache[file_id] = entry
1889
new_path = self.id2path(file_id)
1890
else:
1891
entry = None
1892
new_path = None
1893
delta.append((old_path, new_path, file_id, entry))
1894
return delta
1895
1896
def path2id(self, name):
1897
"""See CommonInventory.path2id()."""
1898
result = self._path_to_fileid_cache.get(name, None)
1899
if result is None:
1900
result = CommonInventory.path2id(self, name)
1901
self._path_to_fileid_cache[name] = result
1902
return result
1903
1904
def to_lines(self):
1905
"""Serialise the inventory to lines."""
1906
lines = ["chkinventory:\n"]
1907
if self._search_key_name != 'plain':
1908
# custom ordering grouping things that don't change together
1909
lines.append('search_key_name: %s\n' % (self._search_key_name,))
1910
lines.append("root_id: %s\n" % self.root_id)
1911
lines.append('parent_id_basename_to_file_id: %s\n' %
1912
self.parent_id_basename_to_file_id.key())
1913
lines.append("revision_id: %s\n" % self.revision_id)
1914
lines.append("id_to_entry: %s\n" % self.id_to_entry.key())
1915
else:
1916
lines.append("revision_id: %s\n" % self.revision_id)
1917
lines.append("root_id: %s\n" % self.root_id)
1918
if self.parent_id_basename_to_file_id is not None:
1919
lines.append('parent_id_basename_to_file_id: %s\n' %
1920
self.parent_id_basename_to_file_id.key())
1921
lines.append("id_to_entry: %s\n" % self.id_to_entry.key())
1922
return lines
1923
1924
@property
1925
def root(self):
1926
"""Get the root entry."""
1927
return self[self.root_id]
1928
1929
1930
class CHKInventoryDirectory(InventoryDirectory):
1931
"""A directory in an inventory."""
1932
1933
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
1934
'text_id', 'parent_id', '_children', 'executable',
1935
'revision', 'symlink_target', 'reference_revision',
1936
'_chk_inventory']
1937
1938
def __init__(self, file_id, name, parent_id, chk_inventory):
1939
# Don't call InventoryDirectory.__init__ - it isn't right for this
1940
# class.
1941
InventoryEntry.__init__(self, file_id, name, parent_id)
1942
self._children = None
1943
self.kind = 'directory'
1944
self._chk_inventory = chk_inventory
1945
1946
@property
1947
def children(self):
1948
"""Access the list of children of this directory.
1949
1950
With a parent_id_basename_to_file_id index, loads all the children,
1951
without loads the entire index. Without is bad. A more sophisticated
1952
proxy object might be nice, to allow partial loading of children as
1953
well when specific names are accessed. (So path traversal can be
1954
written in the obvious way but not examine siblings.).
1955
"""
1956
if self._children is not None:
1957
return self._children
1958
# No longer supported
1959
if self._chk_inventory.parent_id_basename_to_file_id is None:
1960
raise AssertionError("Inventories without"
1961
" parent_id_basename_to_file_id are no longer supported")
1962
result = {}
1963
# XXX: Todo - use proxy objects for the children rather than loading
1964
# all when the attribute is referenced.
1965
parent_id_index = self._chk_inventory.parent_id_basename_to_file_id
1966
child_keys = set()
1967
for (parent_id, name_utf8), file_id in parent_id_index.iteritems(
1968
key_filter=[(self.file_id,)]):
1969
child_keys.add((file_id,))
1970
cached = set()
1971
for file_id_key in child_keys:
1972
entry = self._chk_inventory._fileid_to_entry_cache.get(
1973
file_id_key[0], None)
1974
if entry is not None:
1975
result[entry.name] = entry
1976
cached.add(file_id_key)
1977
child_keys.difference_update(cached)
1978
# populate; todo: do by name
1979
id_to_entry = self._chk_inventory.id_to_entry
1980
for file_id_key, bytes in id_to_entry.iteritems(child_keys):
1981
entry = self._chk_inventory._bytes_to_entry(bytes)
1982
result[entry.name] = entry
1983
self._chk_inventory._fileid_to_entry_cache[file_id_key[0]] = entry
1984
self._children = result
1985
return result
1382
1986
1383
1987
entry_factory = {
1384
1988
'directory': InventoryDirectory,
Loggerhead is a web-based interface for Breezy
Version: 2.0.1