300
239
raise BzrError("don't know how to export {%s} of kind %r" %
301
240
(self.file_id, self.kind))
242
@deprecated_method(deprecated_in((1, 6, 0)))
303
243
def put_on_disk(self, dest, dp, tree):
304
244
"""Create a representation of self on disk in the prefix dest.
306
246
This is a template method - implement _put_on_disk in subclasses.
308
fullpath = pathjoin(dest, dp)
248
fullpath = osutils.pathjoin(dest, dp)
309
249
self._put_on_disk(fullpath, tree)
310
mutter(" export {%s} kind %s to %s", self.file_id,
250
# mutter(" export {%s} kind %s to %s", self.file_id,
251
# self.kind, fullpath)
313
253
def _put_on_disk(self, fullpath, tree):
314
254
"""Put this entry onto disk at fullpath, from tree tree."""
315
255
raise BzrError("don't know how to export {%s} of kind %r" % (self.file_id, self.kind))
317
257
def sorted_children(self):
318
l = self.children.items()
258
return sorted(self.children.items())
323
261
def versionable_kind(kind):
324
return kind in ('file', 'directory', 'symlink')
262
return (kind in ('file', 'directory', 'symlink', 'tree-reference'))
326
def check(self, checker, rev_id, inv, tree):
264
def check(self, checker, rev_id, inv):
327
265
"""Check this inventory entry is intact.
329
267
This is a template method, override _check for kind specific
332
:param checker: Check object providing context for the checks;
270
:param checker: Check object providing context for the checks;
333
271
can be used to find out what parts of the repository have already
335
273
:param rev_id: Revision id from which this InventoryEntry was loaded.
336
274
Not necessarily the last-changed revision for this file.
337
275
:param inv: Inventory from which the entry was loaded.
338
:param tree: RevisionTree for this entry.
340
if self.parent_id != None:
277
if self.parent_id is not None:
341
278
if not inv.has_id(self.parent_id):
342
279
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
343
280
% (self.parent_id, rev_id))
344
self._check(checker, rev_id, tree)
281
checker._add_entry_to_text_key_references(inv, self)
282
self._check(checker, rev_id)
346
def _check(self, checker, rev_id, tree):
284
def _check(self, checker, rev_id):
347
285
"""Check this inventory entry for kind specific errors."""
348
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
286
checker._report_items.append(
287
'unknown entry kind %r in revision {%s}' % (self.kind, rev_id))
353
290
"""Clone this inventory entry."""
354
291
raise NotImplementedError
356
def _get_snapshot_change(self, previous_entries):
357
if len(previous_entries) > 1:
359
elif len(previous_entries) == 0:
294
def describe_change(old_entry, new_entry):
295
"""Describe the change between old_entry and this.
297
This smells of being an InterInventoryEntry situation, but as its
298
the first one, we're making it a static method for now.
300
An entry with a different parent, or different name is considered
301
to be renamed. Reparenting is an internal detail.
302
Note that renaming the parent does not trigger a rename for the
305
# TODO: Perhaps return an object rather than just a string
306
if old_entry is new_entry:
307
# also the case of both being None
309
elif old_entry is None:
362
return 'modified/renamed/reparented'
311
elif new_entry is None:
313
if old_entry.kind != new_entry.kind:
315
text_modified, meta_modified = new_entry.detect_changes(old_entry)
316
if text_modified or meta_modified:
320
# TODO 20060511 (mbp, rbc) factor out 'detect_rename' here.
321
if old_entry.parent_id != new_entry.parent_id:
323
elif old_entry.name != new_entry.name:
327
if renamed and not modified:
328
return InventoryEntry.RENAMED
329
if modified and not renamed:
331
if modified and renamed:
332
return InventoryEntry.MODIFIED_AND_RENAMED
364
335
def __repr__(self):
365
return ("%s(%r, %r, parent_id=%r)"
336
return ("%s(%r, %r, parent_id=%r, revision=%r)"
366
337
% (self.__class__.__name__,
371
def snapshot(self, revision, path, previous_entries,
372
work_tree, weave_store, transaction):
373
"""Make a snapshot of this entry which may or may not have changed.
375
This means that all its fields are populated, that it has its
376
text stored in the text store or weave.
378
mutter('new parents of %s are %r', path, previous_entries)
379
self._read_tree_state(path, work_tree)
380
if len(previous_entries) == 1:
381
# cannot be unchanged unless there is only one parent file rev.
382
parent_ie = previous_entries.values()[0]
383
if self._unchanged(parent_ie):
384
mutter("found unchanged entry")
385
self.revision = parent_ie.revision
387
return self.snapshot_revision(revision, previous_entries,
388
work_tree, weave_store, transaction)
390
def snapshot_revision(self, revision, previous_entries, work_tree,
391
weave_store, transaction):
392
"""Record this revision unconditionally."""
393
mutter('new revision for {%s}', self.file_id)
394
self.revision = revision
395
change = self._get_snapshot_change(previous_entries)
396
self._snapshot_text(previous_entries, work_tree, weave_store,
400
def _snapshot_text(self, file_parents, work_tree, weave_store, transaction):
401
"""Record the 'text' of this entry, whatever form that takes.
403
This default implementation simply adds an empty text.
405
mutter('storing file {%s} in revision {%s}',
406
self.file_id, self.revision)
407
self._add_text_to_weave([], file_parents.keys(), weave_store, transaction)
409
343
def __eq__(self, other):
345
# For the case when objects are cached
410
347
if not isinstance(other, InventoryEntry):
411
348
return NotImplemented
741
687
return compatible
744
class Inventory(object):
745
"""Inventory of versioned files in a tree.
747
This describes which file_id is present at each point in the tree,
748
and possibly the SHA-1 or other information about the file.
690
class TreeReference(InventoryEntry):
692
kind = 'tree-reference'
694
def __init__(self, file_id, name, parent_id, revision=None,
695
reference_revision=None):
696
InventoryEntry.__init__(self, file_id, name, parent_id)
697
self.revision = revision
698
self.reference_revision = reference_revision
701
return TreeReference(self.file_id, self.name, self.parent_id,
702
self.revision, self.reference_revision)
704
def _read_tree_state(self, path, work_tree):
705
"""Populate fields in the inventory entry from the given tree.
707
self.reference_revision = work_tree.get_reference_revision(
710
def _forget_tree_state(self):
711
self.reference_revision = None
713
def _unchanged(self, previous_ie):
714
"""See InventoryEntry._unchanged."""
715
compatible = super(TreeReference, self)._unchanged(previous_ie)
716
if self.reference_revision != previous_ie.reference_revision:
721
class CommonInventory(object):
722
"""Basic inventory logic, defined in terms of primitives like has_id.
724
An inventory is the metadata about the contents of a tree.
726
This is broadly a map from file_id to entries such as directories, files,
727
symlinks and tree references. Each entry maintains its own metadata like
728
SHA1 and length for files, or children for a directory.
749
730
Entries can be looked up either by path or by file_id.
751
The inventory represents a typical unix file tree, with
752
directories containing files and subdirectories. We never store
753
the full path to a file, because renaming a directory implicitly
754
moves all of its contents. This class internally maintains a
732
InventoryEntry objects must not be modified after they are
733
inserted, other than through the Inventory API.
736
def __contains__(self, file_id):
737
"""True if this entry contains a file with given id.
739
>>> inv = Inventory()
740
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
741
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
747
Note that this method along with __iter__ are not encouraged for use as
748
they are less clear than specific query methods - they may be rmeoved
751
return self.has_id(file_id)
753
def has_filename(self, filename):
754
return bool(self.path2id(filename))
756
def id2path(self, file_id):
757
"""Return as a string the path to file_id.
760
>>> e = i.add(InventoryDirectory('src-id', 'src', ROOT_ID))
761
>>> e = i.add(InventoryFile('foo-id', 'foo.c', parent_id='src-id'))
762
>>> print i.id2path('foo-id')
765
:raises NoSuchId: If file_id is not present in the inventory.
767
# get all names, skipping root
768
return '/'.join(reversed(
769
[parent.name for parent in
770
self._iter_file_id_parents(file_id)][:-1]))
772
def iter_entries(self, from_dir=None, recursive=True):
773
"""Return (path, entry) pairs, in order by name.
775
:param from_dir: if None, start from the root,
776
otherwise start from this directory (either file-id or entry)
777
:param recursive: recurse into directories or not
780
if self.root is None:
784
elif isinstance(from_dir, basestring):
785
from_dir = self[from_dir]
787
# unrolling the recursive called changed the time from
788
# 440ms/663ms (inline/total) to 116ms/116ms
789
children = from_dir.children.items()
792
for name, ie in children:
795
children = collections.deque(children)
796
stack = [(u'', children)]
798
from_dir_relpath, children = stack[-1]
801
name, ie = children.popleft()
803
# we know that from_dir_relpath never ends in a slash
804
# and 'f' doesn't begin with one, we can do a string op, rather
805
# than the checks of pathjoin(), though this means that all paths
807
path = from_dir_relpath + '/' + name
811
if ie.kind != 'directory':
814
# But do this child first
815
new_children = ie.children.items()
817
new_children = collections.deque(new_children)
818
stack.append((path, new_children))
819
# Break out of inner loop, so that we start outer loop with child
822
# if we finished all children, pop it off the stack
825
def iter_entries_by_dir(self, from_dir=None, specific_file_ids=None,
826
yield_parents=False):
827
"""Iterate over the entries in a directory first order.
829
This returns all entries for a directory before returning
830
the entries for children of a directory. This is not
831
lexicographically sorted order, and is a hybrid between
832
depth-first and breadth-first.
834
:param yield_parents: If True, yield the parents from the root leading
835
down to specific_file_ids that have been requested. This has no
836
impact if specific_file_ids is None.
837
:return: This yields (path, entry) pairs
839
if specific_file_ids and not isinstance(specific_file_ids, set):
840
specific_file_ids = set(specific_file_ids)
841
# TODO? Perhaps this should return the from_dir so that the root is
842
# yielded? or maybe an option?
844
if self.root is None:
846
# Optimize a common case
847
if (not yield_parents and specific_file_ids is not None and
848
len(specific_file_ids) == 1):
849
file_id = list(specific_file_ids)[0]
851
yield self.id2path(file_id), self[file_id]
854
if (specific_file_ids is None or yield_parents or
855
self.root.file_id in specific_file_ids):
857
elif isinstance(from_dir, basestring):
858
from_dir = self[from_dir]
860
if specific_file_ids is not None:
861
# TODO: jam 20070302 This could really be done as a loop rather
862
# than a bunch of recursive calls.
865
def add_ancestors(file_id):
866
if file_id not in byid:
868
parent_id = byid[file_id].parent_id
869
if parent_id is None:
871
if parent_id not in parents:
872
parents.add(parent_id)
873
add_ancestors(parent_id)
874
for file_id in specific_file_ids:
875
add_ancestors(file_id)
879
stack = [(u'', from_dir)]
881
cur_relpath, cur_dir = stack.pop()
884
for child_name, child_ie in sorted(cur_dir.children.iteritems()):
886
child_relpath = cur_relpath + child_name
888
if (specific_file_ids is None or
889
child_ie.file_id in specific_file_ids or
890
(yield_parents and child_ie.file_id in parents)):
891
yield child_relpath, child_ie
893
if child_ie.kind == 'directory':
894
if parents is None or child_ie.file_id in parents:
895
child_dirs.append((child_relpath+'/', child_ie))
896
stack.extend(reversed(child_dirs))
898
def _make_delta(self, old):
899
"""Make an inventory delta from two inventories."""
902
adds = new_ids - old_ids
903
deletes = old_ids - new_ids
904
common = old_ids.intersection(new_ids)
906
for file_id in deletes:
907
delta.append((old.id2path(file_id), None, file_id, None))
909
delta.append((None, self.id2path(file_id), file_id, self[file_id]))
910
for file_id in common:
911
if old[file_id] != self[file_id]:
912
delta.append((old.id2path(file_id), self.id2path(file_id),
913
file_id, self[file_id]))
916
def _get_mutable_inventory(self):
917
"""Returns a mutable copy of the object.
919
Some inventories are immutable, yet working trees, for example, needs
920
to mutate exisiting inventories instead of creating a new one.
922
raise NotImplementedError(self._get_mutable_inventory)
924
def make_entry(self, kind, name, parent_id, file_id=None):
925
"""Simple thunk to bzrlib.inventory.make_entry."""
926
return make_entry(kind, name, parent_id, file_id)
929
"""Return list of (path, ie) for all entries except the root.
931
This may be faster than iter_entries.
934
def descend(dir_ie, dir_path):
935
kids = dir_ie.children.items()
937
for name, ie in kids:
938
child_path = osutils.pathjoin(dir_path, name)
939
accum.append((child_path, ie))
940
if ie.kind == 'directory':
941
descend(ie, child_path)
943
descend(self.root, u'')
946
def directories(self):
947
"""Return (path, entry) pairs for all directories, including the root.
950
def descend(parent_ie, parent_path):
951
accum.append((parent_path, parent_ie))
953
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
956
for name, child_ie in kids:
957
child_path = osutils.pathjoin(parent_path, name)
958
descend(child_ie, child_path)
959
descend(self.root, u'')
962
def path2id(self, relpath):
963
"""Walk down through directories to return entry of last component.
965
:param relpath: may be either a list of path components, or a single
966
string, in which case it is automatically split.
968
This returns the entry of the last component in the path,
969
which may be either a file or a directory.
971
Returns None IFF the path is not found.
973
if isinstance(relpath, basestring):
974
names = osutils.splitpath(relpath)
980
except errors.NoSuchId:
981
# root doesn't exist yet so nothing else can
987
children = getattr(parent, 'children', None)
996
return parent.file_id
998
def filter(self, specific_fileids):
999
"""Get an inventory view filtered against a set of file-ids.
1001
Children of directories and parents are included.
1003
The result may or may not reference the underlying inventory
1004
so it should be treated as immutable.
1006
interesting_parents = set()
1007
for fileid in specific_fileids:
1009
interesting_parents.update(self.get_idpath(fileid))
1010
except errors.NoSuchId:
1011
# This fileid is not in the inventory - that's ok
1013
entries = self.iter_entries()
1014
if self.root is None:
1015
return Inventory(root_id=None)
1016
other = Inventory(entries.next()[1].file_id)
1017
other.root.revision = self.root.revision
1018
other.revision_id = self.revision_id
1019
directories_to_expand = set()
1020
for path, entry in entries:
1021
file_id = entry.file_id
1022
if (file_id in specific_fileids
1023
or entry.parent_id in directories_to_expand):
1024
if entry.kind == 'directory':
1025
directories_to_expand.add(file_id)
1026
elif file_id not in interesting_parents:
1028
other.add(entry.copy())
1031
def get_idpath(self, file_id):
1032
"""Return a list of file_ids for the path to an entry.
1034
The list contains one element for each directory followed by
1035
the id of the file itself. So the length of the returned list
1036
is equal to the depth of the file in the tree, counting the
1037
root directory as depth 1.
1040
for parent in self._iter_file_id_parents(file_id):
1041
p.insert(0, parent.file_id)
1045
class Inventory(CommonInventory):
1046
"""Mutable dict based in-memory inventory.
1048
We never store the full path to a file, because renaming a directory
1049
implicitly moves all of its contents. This class internally maintains a
755
1050
lookup tree that allows the children under a directory to be
756
1051
returned quickly.
758
InventoryEntry objects must not be modified after they are
759
inserted, other than through the Inventory API.
761
1053
>>> inv = Inventory()
762
1054
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
763
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT')
1055
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
764
1056
>>> inv['123-123'].name
767
May be treated as an iterator or set to look up file ids:
769
>>> bool(inv.path2id('hello.c'))
1059
Id's may be looked up from paths:
1061
>>> inv.path2id('hello.c')
771
1063
>>> '123-123' in inv
774
May also look up by name:
1066
There are iterators over the contents:
776
>>> [x[0] for x in inv.iter_entries()]
778
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
779
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
780
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT-12345678-12345678')
1068
>>> [entry[0] for entry in inv.iter_entries()]
782
1072
def __init__(self, root_id=ROOT_ID, revision_id=None):
783
1073
"""Create or read an inventory.
789
1079
The inventory is created with a default root directory, with
792
# We are letting Branch.create() create a unique inventory
793
# root id. Rather than generating a random one here.
795
# root_id = bzrlib.branch.gen_file_id('TREE_ROOT')
796
self.root = RootEntry(root_id)
1082
if root_id is not None:
1083
self._set_root(InventoryDirectory(root_id, u'', None))
797
1087
self.revision_id = revision_id
1090
# More than one page of ouput is not useful anymore to debug
1093
contents = repr(self._byid)
1094
if len(contents) > max_len:
1095
contents = contents[:(max_len-len(closing))] + closing
1096
return "<Inventory object at %x, contents=%r>" % (id(self), contents)
1098
def apply_delta(self, delta):
1099
"""Apply a delta to this inventory.
1101
See the inventory developers documentation for the theory behind
1104
If delta application fails the inventory is left in an indeterminate
1105
state and must not be used.
1107
:param delta: A list of changes to apply. After all the changes are
1108
applied the final inventory must be internally consistent, but it
1109
is ok to supply changes which, if only half-applied would have an
1110
invalid result - such as supplying two changes which rename two
1111
files, 'A' and 'B' with each other : [('A', 'B', 'A-id', a_entry),
1112
('B', 'A', 'B-id', b_entry)].
1114
Each change is a tuple, of the form (old_path, new_path, file_id,
1117
When new_path is None, the change indicates the removal of an entry
1118
from the inventory and new_entry will be ignored (using None is
1119
appropriate). If new_path is not None, then new_entry must be an
1120
InventoryEntry instance, which will be incorporated into the
1121
inventory (and replace any existing entry with the same file id).
1123
When old_path is None, the change indicates the addition of
1124
a new entry to the inventory.
1126
When neither new_path nor old_path are None, the change is a
1127
modification to an entry, such as a rename, reparent, kind change
1130
The children attribute of new_entry is ignored. This is because
1131
this method preserves children automatically across alterations to
1132
the parent of the children, and cases where the parent id of a
1133
child is changing require the child to be passed in as a separate
1134
change regardless. E.g. in the recursive deletion of a directory -
1135
the directory's children must be included in the delta, or the
1136
final inventory will be invalid.
1138
Note that a file_id must only appear once within a given delta.
1139
An AssertionError is raised otherwise.
1141
# Check that the delta is legal. It would be nice if this could be
1142
# done within the loops below but it's safer to validate the delta
1143
# before starting to mutate the inventory, as there isn't a rollback
1145
list(_check_delta_unique_ids(_check_delta_unique_new_paths(
1146
_check_delta_unique_old_paths(_check_delta_ids_match_entry(
1147
_check_delta_ids_are_valid(
1148
_check_delta_new_path_entry_both_or_None(
1152
# Remove all affected items which were in the original inventory,
1153
# starting with the longest paths, thus ensuring parents are examined
1154
# after their children, which means that everything we examine has no
1155
# modified children remaining by the time we examine it.
1156
for old_path, file_id in sorted(((op, f) for op, np, f, e in delta
1157
if op is not None), reverse=True):
1158
# Preserve unaltered children of file_id for later reinsertion.
1159
file_id_children = getattr(self[file_id], 'children', {})
1160
if len(file_id_children):
1161
children[file_id] = file_id_children
1162
if self.id2path(file_id) != old_path:
1163
raise errors.InconsistentDelta(old_path, file_id,
1164
"Entry was at wrong other path %r." % self.id2path(file_id))
1165
# Remove file_id and the unaltered children. If file_id is not
1166
# being deleted it will be reinserted back later.
1167
self.remove_recursive_id(file_id)
1168
# Insert all affected which should be in the new inventory, reattaching
1169
# their children if they had any. This is done from shortest path to
1170
# longest, ensuring that items which were modified and whose parents in
1171
# the resulting inventory were also modified, are inserted after their
1173
for new_path, f, new_entry in sorted((np, f, e) for op, np, f, e in
1174
delta if np is not None):
1175
if new_entry.kind == 'directory':
1176
# Pop the child which to allow detection of children whose
1177
# parents were deleted and which were not reattached to a new
1179
replacement = InventoryDirectory(new_entry.file_id,
1180
new_entry.name, new_entry.parent_id)
1181
replacement.revision = new_entry.revision
1182
replacement.children = children.pop(replacement.file_id, {})
1183
new_entry = replacement
1186
except errors.DuplicateFileId:
1187
raise errors.InconsistentDelta(new_path, new_entry.file_id,
1188
"New id is already present in target.")
1189
except AttributeError:
1190
raise errors.InconsistentDelta(new_path, new_entry.file_id,
1191
"Parent is not a directory.")
1192
if self.id2path(new_entry.file_id) != new_path:
1193
raise errors.InconsistentDelta(new_path, new_entry.file_id,
1194
"New path is not consistent with parent path.")
1196
# Get the parent id that was deleted
1197
parent_id, children = children.popitem()
1198
raise errors.InconsistentDelta("<deleted>", parent_id,
1199
"The file id was deleted but its children were not deleted.")
1201
def create_by_apply_delta(self, inventory_delta, new_revision_id,
1202
propagate_caches=False):
1203
"""See CHKInventory.create_by_apply_delta()"""
1204
new_inv = self.copy()
1205
new_inv.apply_delta(inventory_delta)
1206
new_inv.revision_id = new_revision_id
1209
def _set_root(self, ie):
798
1211
self._byid = {self.root.file_id: self.root}
802
1214
# TODO: jam 20051218 Should copy also copy the revision_id?
803
other = Inventory(self.root.file_id)
1215
entries = self.iter_entries()
1216
if self.root is None:
1217
return Inventory(root_id=None)
1218
other = Inventory(entries.next()[1].file_id)
1219
other.root.revision = self.root.revision
804
1220
# copy recursively so we know directories will be added before
805
1221
# their children. There are more efficient ways than this...
806
for path, entry in self.iter_entries():
807
if entry == self.root:
1222
for path, entry in entries:
809
1223
other.add(entry.copy())
1226
def _get_mutable_inventory(self):
1227
"""See CommonInventory._get_mutable_inventory."""
1228
return copy.deepcopy(self)
813
1230
def __iter__(self):
1231
"""Iterate over all file-ids."""
814
1232
return iter(self._byid)
1234
def iter_just_entries(self):
1235
"""Iterate over all entries.
1237
Unlike iter_entries(), just the entries are returned (not (path, ie))
1238
and the order of entries is undefined.
1240
XXX: We may not want to merge this into bzr.dev.
1242
if self.root is None:
1244
for _, ie in self._byid.iteritems():
817
1247
def __len__(self):
818
1248
"""Returns number of entries."""
819
1249
return len(self._byid)
822
def iter_entries(self, from_dir=None):
823
"""Return (path, entry) pairs, in order by name."""
827
elif isinstance(from_dir, basestring):
828
from_dir = self._byid[from_dir]
830
kids = from_dir.children.items()
832
for name, ie in kids:
834
if ie.kind == 'directory':
835
for cn, cie in self.iter_entries(from_dir=ie.file_id):
836
yield pathjoin(name, cn), cie
840
"""Return list of (path, ie) for all entries except the root.
842
This may be faster than iter_entries.
845
def descend(dir_ie, dir_path):
846
kids = dir_ie.children.items()
848
for name, ie in kids:
849
child_path = pathjoin(dir_path, name)
850
accum.append((child_path, ie))
851
if ie.kind == 'directory':
852
descend(ie, child_path)
854
descend(self.root, u'')
858
def directories(self):
859
"""Return (path, entry) pairs for all directories, including the root.
862
def descend(parent_ie, parent_path):
863
accum.append((parent_path, parent_ie))
865
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
868
for name, child_ie in kids:
869
child_path = pathjoin(parent_path, name)
870
descend(child_ie, child_path)
871
descend(self.root, u'')
876
def __contains__(self, file_id):
877
"""True if this entry contains a file with given id.
879
>>> inv = Inventory()
880
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
881
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT')
887
return file_id in self._byid
890
1251
def __getitem__(self, file_id):
891
1252
"""Return the entry for given file_id.
893
1254
>>> inv = Inventory()
894
1255
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
895
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT')
1256
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
896
1257
>>> inv['123123'].name
900
1261
return self._byid[file_id]
901
1262
except KeyError:
903
raise BzrError("can't look up file_id None")
905
raise BzrError("file_id {%s} not in inventory" % file_id)
1263
# really we're passing an inventory, not a tree...
1264
raise errors.NoSuchId(self, file_id)
908
1266
def get_file_kind(self, file_id):
909
1267
return self._byid[file_id].kind
1129
1465
del old_parent.children[file_ie.name]
1130
1466
new_parent.children[new_name] = file_ie
1132
1468
file_ie.name = new_name
1133
1469
file_ie.parent_id = new_parent_id
1471
def is_root(self, file_id):
1472
return self.root is not None and file_id == self.root.file_id
1475
class CHKInventory(CommonInventory):
1476
"""An inventory persisted in a CHK store.
1478
By design, a CHKInventory is immutable so many of the methods
1479
supported by Inventory - add, rename, apply_delta, etc - are *not*
1480
supported. To create a new CHKInventory, use create_by_apply_delta()
1481
or from_inventory(), say.
1483
Internally, a CHKInventory has one or two CHKMaps:
1485
* id_to_entry - a map from (file_id,) => InventoryEntry as bytes
1486
* parent_id_basename_to_file_id - a map from (parent_id, basename_utf8)
1489
The second map is optional and not present in early CHkRepository's.
1491
No caching is performed: every method call or item access will perform
1492
requests to the storage layer. As such, keep references to objects you
1496
def __init__(self, search_key_name):
1497
CommonInventory.__init__(self)
1498
self._fileid_to_entry_cache = {}
1499
self._path_to_fileid_cache = {}
1500
self._search_key_name = search_key_name
1503
def __eq__(self, other):
1504
"""Compare two sets by comparing their contents."""
1505
if not isinstance(other, CHKInventory):
1506
return NotImplemented
1508
this_key = self.id_to_entry.key()
1509
other_key = other.id_to_entry.key()
1510
this_pid_key = self.parent_id_basename_to_file_id.key()
1511
other_pid_key = other.parent_id_basename_to_file_id.key()
1512
if None in (this_key, this_pid_key, other_key, other_pid_key):
1514
return this_key == other_key and this_pid_key == other_pid_key
1516
def _entry_to_bytes(self, entry):
1517
"""Serialise entry as a single bytestring.
1519
:param Entry: An inventory entry.
1520
:return: A bytestring for the entry.
1523
ENTRY ::= FILE | DIR | SYMLINK | TREE
1524
FILE ::= "file: " COMMON SEP SHA SEP SIZE SEP EXECUTABLE
1525
DIR ::= "dir: " COMMON
1526
SYMLINK ::= "symlink: " COMMON SEP TARGET_UTF8
1527
TREE ::= "tree: " COMMON REFERENCE_REVISION
1528
COMMON ::= FILE_ID SEP PARENT_ID SEP NAME_UTF8 SEP REVISION
1531
if entry.parent_id is not None:
1532
parent_str = entry.parent_id
1535
name_str = entry.name.encode("utf8")
1536
if entry.kind == 'file':
1537
if entry.executable:
1541
return "file: %s\n%s\n%s\n%s\n%s\n%d\n%s" % (
1542
entry.file_id, parent_str, name_str, entry.revision,
1543
entry.text_sha1, entry.text_size, exec_str)
1544
elif entry.kind == 'directory':
1545
return "dir: %s\n%s\n%s\n%s" % (
1546
entry.file_id, parent_str, name_str, entry.revision)
1547
elif entry.kind == 'symlink':
1548
return "symlink: %s\n%s\n%s\n%s\n%s" % (
1549
entry.file_id, parent_str, name_str, entry.revision,
1550
entry.symlink_target.encode("utf8"))
1551
elif entry.kind == 'tree-reference':
1552
return "tree: %s\n%s\n%s\n%s\n%s" % (
1553
entry.file_id, parent_str, name_str, entry.revision,
1554
entry.reference_revision)
1556
raise ValueError("unknown kind %r" % entry.kind)
1558
def _expand_fileids_to_parents_and_children(self, file_ids):
1559
"""Give a more wholistic view starting with the given file_ids.
1561
For any file_id which maps to a directory, we will include all children
1562
of that directory. We will also include all directories which are
1563
parents of the given file_ids, but we will not include their children.
1570
fringle # fringle-id
1574
if given [foo-id] we will include
1575
TREE_ROOT as interesting parents
1577
foo-id, baz-id, frob-id, fringle-id
1581
# TODO: Pre-pass over the list of fileids to see if anything is already
1582
# deserialized in self._fileid_to_entry_cache
1584
directories_to_expand = set()
1585
children_of_parent_id = {}
1586
# It is okay if some of the fileids are missing
1587
for entry in self._getitems(file_ids):
1588
if entry.kind == 'directory':
1589
directories_to_expand.add(entry.file_id)
1590
interesting.add(entry.parent_id)
1591
children_of_parent_id.setdefault(entry.parent_id, []
1592
).append(entry.file_id)
1594
# Now, interesting has all of the direct parents, but not the
1595
# parents of those parents. It also may have some duplicates with
1597
remaining_parents = interesting.difference(file_ids)
1598
# When we hit the TREE_ROOT, we'll get an interesting parent of None,
1599
# but we don't actually want to recurse into that
1600
interesting.add(None) # this will auto-filter it in the loop
1601
remaining_parents.discard(None)
1602
while remaining_parents:
1603
next_parents = set()
1604
for entry in self._getitems(remaining_parents):
1605
next_parents.add(entry.parent_id)
1606
children_of_parent_id.setdefault(entry.parent_id, []
1607
).append(entry.file_id)
1608
# Remove any search tips we've already processed
1609
remaining_parents = next_parents.difference(interesting)
1610
interesting.update(remaining_parents)
1611
# We should probably also .difference(directories_to_expand)
1612
interesting.update(file_ids)
1613
interesting.discard(None)
1614
while directories_to_expand:
1615
# Expand directories by looking in the
1616
# parent_id_basename_to_file_id map
1617
keys = [StaticTuple(f,).intern() for f in directories_to_expand]
1618
directories_to_expand = set()
1619
items = self.parent_id_basename_to_file_id.iteritems(keys)
1620
next_file_ids = set([item[1] for item in items])
1621
next_file_ids = next_file_ids.difference(interesting)
1622
interesting.update(next_file_ids)
1623
for entry in self._getitems(next_file_ids):
1624
if entry.kind == 'directory':
1625
directories_to_expand.add(entry.file_id)
1626
children_of_parent_id.setdefault(entry.parent_id, []
1627
).append(entry.file_id)
1628
return interesting, children_of_parent_id
1630
def filter(self, specific_fileids):
1631
"""Get an inventory view filtered against a set of file-ids.
1633
Children of directories and parents are included.
1635
The result may or may not reference the underlying inventory
1636
so it should be treated as immutable.
1639
parent_to_children) = self._expand_fileids_to_parents_and_children(
1641
# There is some overlap here, but we assume that all interesting items
1642
# are in the _fileid_to_entry_cache because we had to read them to
1643
# determine if they were a dir we wanted to recurse, or just a file
1644
# This should give us all the entries we'll want to add, so start
1646
other = Inventory(self.root_id)
1647
other.root.revision = self.root.revision
1648
other.revision_id = self.revision_id
1649
if not interesting or not parent_to_children:
1650
# empty filter, or filtering entrys that don't exist
1651
# (if even 1 existed, then we would have populated
1652
# parent_to_children with at least the tree root.)
1654
cache = self._fileid_to_entry_cache
1656
remaining_children = collections.deque(parent_to_children[self.root_id])
1658
import pdb; pdb.set_trace()
1660
while remaining_children:
1661
file_id = remaining_children.popleft()
1663
if ie.kind == 'directory':
1664
ie = ie.copy() # We create a copy to depopulate the .children attribute
1665
# TODO: depending on the uses of 'other' we should probably alwyas
1666
# '.copy()' to prevent someone from mutating other and
1667
# invaliding our internal cache
1669
if file_id in parent_to_children:
1670
remaining_children.extend(parent_to_children[file_id])
1674
def _bytes_to_utf8name_key(bytes):
1675
"""Get the file_id, revision_id key out of bytes."""
1676
# We don't normally care about name, except for times when we want
1677
# to filter out empty names because of non rich-root...
1678
sections = bytes.split('\n')
1679
kind, file_id = sections[0].split(': ')
1680
return (sections[2], file_id, sections[3])
1682
def _bytes_to_entry(self, bytes):
1683
"""Deserialise a serialised entry."""
1684
sections = bytes.split('\n')
1685
if sections[0].startswith("file: "):
1686
result = InventoryFile(sections[0][6:],
1687
sections[2].decode('utf8'),
1689
result.text_sha1 = sections[4]
1690
result.text_size = int(sections[5])
1691
result.executable = sections[6] == "Y"
1692
elif sections[0].startswith("dir: "):
1693
result = CHKInventoryDirectory(sections[0][5:],
1694
sections[2].decode('utf8'),
1696
elif sections[0].startswith("symlink: "):
1697
result = InventoryLink(sections[0][9:],
1698
sections[2].decode('utf8'),
1700
result.symlink_target = sections[4].decode('utf8')
1701
elif sections[0].startswith("tree: "):
1702
result = TreeReference(sections[0][6:],
1703
sections[2].decode('utf8'),
1705
result.reference_revision = sections[4]
1707
raise ValueError("Not a serialised entry %r" % bytes)
1708
result.revision = sections[3]
1709
if result.parent_id == '':
1710
result.parent_id = None
1711
self._fileid_to_entry_cache[result.file_id] = result
1714
def _get_mutable_inventory(self):
1715
"""See CommonInventory._get_mutable_inventory."""
1716
entries = self.iter_entries()
1717
inv = Inventory(None, self.revision_id)
1718
for path, inv_entry in entries:
1719
inv.add(inv_entry.copy())
1722
def create_by_apply_delta(self, inventory_delta, new_revision_id,
1723
propagate_caches=False):
1724
"""Create a new CHKInventory by applying inventory_delta to this one.
1726
See the inventory developers documentation for the theory behind
1729
:param inventory_delta: The inventory delta to apply. See
1730
Inventory.apply_delta for details.
1731
:param new_revision_id: The revision id of the resulting CHKInventory.
1732
:param propagate_caches: If True, the caches for this inventory are
1733
copied to and updated for the result.
1734
:return: The new CHKInventory.
1736
split = osutils.split
1737
result = CHKInventory(self._search_key_name)
1738
if propagate_caches:
1739
# Just propagate the path-to-fileid cache for now
1740
result._path_to_fileid_cache = dict(self._path_to_fileid_cache.iteritems())
1741
search_key_func = chk_map.search_key_registry.get(self._search_key_name)
1742
self.id_to_entry._ensure_root()
1743
maximum_size = self.id_to_entry._root_node.maximum_size
1744
result.revision_id = new_revision_id
1745
result.id_to_entry = chk_map.CHKMap(
1746
self.id_to_entry._store,
1747
self.id_to_entry.key(),
1748
search_key_func=search_key_func)
1749
result.id_to_entry._ensure_root()
1750
result.id_to_entry._root_node.set_maximum_size(maximum_size)
1751
# Change to apply to the parent_id_basename delta. The dict maps
1752
# (parent_id, basename) -> (old_key, new_value). We use a dict because
1753
# when a path has its id replaced (e.g. the root is changed, or someone
1754
# does bzr mv a b, bzr mv c a, we should output a single change to this
1755
# map rather than two.
1756
parent_id_basename_delta = {}
1757
if self.parent_id_basename_to_file_id is not None:
1758
result.parent_id_basename_to_file_id = chk_map.CHKMap(
1759
self.parent_id_basename_to_file_id._store,
1760
self.parent_id_basename_to_file_id.key(),
1761
search_key_func=search_key_func)
1762
result.parent_id_basename_to_file_id._ensure_root()
1763
self.parent_id_basename_to_file_id._ensure_root()
1764
result_p_id_root = result.parent_id_basename_to_file_id._root_node
1765
p_id_root = self.parent_id_basename_to_file_id._root_node
1766
result_p_id_root.set_maximum_size(p_id_root.maximum_size)
1767
result_p_id_root._key_width = p_id_root._key_width
1769
result.parent_id_basename_to_file_id = None
1770
result.root_id = self.root_id
1771
id_to_entry_delta = []
1772
# inventory_delta is only traversed once, so we just update the
1774
# Check for repeated file ids
1775
inventory_delta = _check_delta_unique_ids(inventory_delta)
1776
# Repeated old paths
1777
inventory_delta = _check_delta_unique_old_paths(inventory_delta)
1778
# Check for repeated new paths
1779
inventory_delta = _check_delta_unique_new_paths(inventory_delta)
1780
# Check for entries that don't match the fileid
1781
inventory_delta = _check_delta_ids_match_entry(inventory_delta)
1782
# Check for nonsense fileids
1783
inventory_delta = _check_delta_ids_are_valid(inventory_delta)
1784
# Check for new_path <-> entry consistency
1785
inventory_delta = _check_delta_new_path_entry_both_or_None(
1787
# All changed entries need to have their parents be directories and be
1788
# at the right path. This set contains (path, id) tuples.
1790
# When we delete an item, all the children of it must be either deleted
1791
# or altered in their own right. As we batch process the change via
1792
# CHKMap.apply_delta, we build a set of things to use to validate the
1796
for old_path, new_path, file_id, entry in inventory_delta:
1799
result.root_id = file_id
1800
if new_path is None:
1805
if propagate_caches:
1807
del result._path_to_fileid_cache[old_path]
1810
deletes.add(file_id)
1812
new_key = StaticTuple(file_id,)
1813
new_value = result._entry_to_bytes(entry)
1814
# Update caches. It's worth doing this whether
1815
# we're propagating the old caches or not.
1816
result._path_to_fileid_cache[new_path] = file_id
1817
parents.add((split(new_path)[0], entry.parent_id))
1818
if old_path is None:
1821
old_key = StaticTuple(file_id,)
1822
if self.id2path(file_id) != old_path:
1823
raise errors.InconsistentDelta(old_path, file_id,
1824
"Entry was at wrong other path %r." %
1825
self.id2path(file_id))
1826
altered.add(file_id)
1827
id_to_entry_delta.append(StaticTuple(old_key, new_key, new_value))
1828
if result.parent_id_basename_to_file_id is not None:
1829
# parent_id, basename changes
1830
if old_path is None:
1833
old_entry = self[file_id]
1834
old_key = self._parent_id_basename_key(old_entry)
1835
if new_path is None:
1839
new_key = self._parent_id_basename_key(entry)
1841
# If the two keys are the same, the value will be unchanged
1842
# as its always the file id for this entry.
1843
if old_key != new_key:
1844
# Transform a change into explicit delete/add preserving
1845
# a possible match on the key from a different file id.
1846
if old_key is not None:
1847
parent_id_basename_delta.setdefault(
1848
old_key, [None, None])[0] = old_key
1849
if new_key is not None:
1850
parent_id_basename_delta.setdefault(
1851
new_key, [None, None])[1] = new_value
1852
# validate that deletes are complete.
1853
for file_id in deletes:
1854
entry = self[file_id]
1855
if entry.kind != 'directory':
1857
# This loop could potentially be better by using the id_basename
1858
# map to just get the child file ids.
1859
for child in entry.children.values():
1860
if child.file_id not in altered:
1861
raise errors.InconsistentDelta(self.id2path(child.file_id),
1862
child.file_id, "Child not deleted or reparented when "
1864
result.id_to_entry.apply_delta(id_to_entry_delta)
1865
if parent_id_basename_delta:
1866
# Transform the parent_id_basename delta data into a linear delta
1867
# with only one record for a given key. Optimally this would allow
1868
# re-keying, but its simpler to just output that as a delete+add
1869
# to spend less time calculating the delta.
1871
for key, (old_key, value) in parent_id_basename_delta.iteritems():
1872
if value is not None:
1873
delta_list.append((old_key, key, value))
1875
delta_list.append((old_key, None, None))
1876
result.parent_id_basename_to_file_id.apply_delta(delta_list)
1877
parents.discard(('', None))
1878
for parent_path, parent in parents:
1880
if result[parent].kind != 'directory':
1881
raise errors.InconsistentDelta(result.id2path(parent), parent,
1882
'Not a directory, but given children')
1883
except errors.NoSuchId:
1884
raise errors.InconsistentDelta("<unknown>", parent,
1885
"Parent is not present in resulting inventory.")
1886
if result.path2id(parent_path) != parent:
1887
raise errors.InconsistentDelta(parent_path, parent,
1888
"Parent has wrong path %r." % result.path2id(parent_path))
1892
def deserialise(klass, chk_store, bytes, expected_revision_id):
1893
"""Deserialise a CHKInventory.
1895
:param chk_store: A CHK capable VersionedFiles instance.
1896
:param bytes: The serialised bytes.
1897
:param expected_revision_id: The revision ID we think this inventory is
1899
:return: A CHKInventory
1901
lines = bytes.split('\n')
1903
raise AssertionError('bytes to deserialize must end with an eol')
1905
if lines[0] != 'chkinventory:':
1906
raise ValueError("not a serialised CHKInventory: %r" % bytes)
1908
allowed_keys = frozenset(['root_id', 'revision_id', 'search_key_name',
1909
'parent_id_basename_to_file_id',
1911
for line in lines[1:]:
1912
key, value = line.split(': ', 1)
1913
if key not in allowed_keys:
1914
raise errors.BzrError('Unknown key in inventory: %r\n%r'
1917
raise errors.BzrError('Duplicate key in inventory: %r\n%r'
1920
revision_id = intern(info['revision_id'])
1921
root_id = intern(info['root_id'])
1922
search_key_name = intern(info.get('search_key_name', 'plain'))
1923
parent_id_basename_to_file_id = intern(info.get(
1924
'parent_id_basename_to_file_id', None))
1925
if not parent_id_basename_to_file_id.startswith('sha1:'):
1926
raise ValueError('parent_id_basename_to_file_id should be a sha1'
1927
' key not %r' % (parent_id_basename_to_file_id,))
1928
id_to_entry = info['id_to_entry']
1929
if not id_to_entry.startswith('sha1:'):
1930
raise ValueError('id_to_entry should be a sha1'
1931
' key not %r' % (id_to_entry,))
1933
result = CHKInventory(search_key_name)
1934
result.revision_id = revision_id
1935
result.root_id = root_id
1936
search_key_func = chk_map.search_key_registry.get(
1937
result._search_key_name)
1938
if parent_id_basename_to_file_id is not None:
1939
result.parent_id_basename_to_file_id = chk_map.CHKMap(
1940
chk_store, StaticTuple(parent_id_basename_to_file_id,),
1941
search_key_func=search_key_func)
1943
result.parent_id_basename_to_file_id = None
1945
result.id_to_entry = chk_map.CHKMap(chk_store,
1946
StaticTuple(id_to_entry,),
1947
search_key_func=search_key_func)
1948
if (result.revision_id,) != expected_revision_id:
1949
raise ValueError("Mismatched revision id and expected: %r, %r" %
1950
(result.revision_id, expected_revision_id))
1954
def from_inventory(klass, chk_store, inventory, maximum_size=0, search_key_name='plain'):
1955
"""Create a CHKInventory from an existing inventory.
1957
The content of inventory is copied into the chk_store, and a
1958
CHKInventory referencing that is returned.
1960
:param chk_store: A CHK capable VersionedFiles instance.
1961
:param inventory: The inventory to copy.
1962
:param maximum_size: The CHKMap node size limit.
1963
:param search_key_name: The identifier for the search key function
1965
result = klass(search_key_name)
1966
result.revision_id = inventory.revision_id
1967
result.root_id = inventory.root.file_id
1969
entry_to_bytes = result._entry_to_bytes
1970
parent_id_basename_key = result._parent_id_basename_key
1971
id_to_entry_dict = {}
1972
parent_id_basename_dict = {}
1973
for path, entry in inventory.iter_entries():
1974
key = StaticTuple(entry.file_id,).intern()
1975
id_to_entry_dict[key] = entry_to_bytes(entry)
1976
p_id_key = parent_id_basename_key(entry)
1977
parent_id_basename_dict[p_id_key] = entry.file_id
1979
result._populate_from_dicts(chk_store, id_to_entry_dict,
1980
parent_id_basename_dict, maximum_size=maximum_size)
1983
def _populate_from_dicts(self, chk_store, id_to_entry_dict,
1984
parent_id_basename_dict, maximum_size):
1985
search_key_func = chk_map.search_key_registry.get(self._search_key_name)
1986
root_key = chk_map.CHKMap.from_dict(chk_store, id_to_entry_dict,
1987
maximum_size=maximum_size, key_width=1,
1988
search_key_func=search_key_func)
1989
self.id_to_entry = chk_map.CHKMap(chk_store, root_key,
1991
root_key = chk_map.CHKMap.from_dict(chk_store,
1992
parent_id_basename_dict,
1993
maximum_size=maximum_size, key_width=2,
1994
search_key_func=search_key_func)
1995
self.parent_id_basename_to_file_id = chk_map.CHKMap(chk_store,
1996
root_key, search_key_func)
1998
def _parent_id_basename_key(self, entry):
1999
"""Create a key for a entry in a parent_id_basename_to_file_id index."""
2000
if entry.parent_id is not None:
2001
parent_id = entry.parent_id
2004
return StaticTuple(parent_id, entry.name.encode('utf8')).intern()
2006
def __getitem__(self, file_id):
2007
"""map a single file_id -> InventoryEntry."""
2009
raise errors.NoSuchId(self, file_id)
2010
result = self._fileid_to_entry_cache.get(file_id, None)
2011
if result is not None:
2014
return self._bytes_to_entry(
2015
self.id_to_entry.iteritems([StaticTuple(file_id,)]).next()[1])
2016
except StopIteration:
2017
# really we're passing an inventory, not a tree...
2018
raise errors.NoSuchId(self, file_id)
2020
def _getitems(self, file_ids):
2021
"""Similar to __getitem__, but lets you query for multiple.
2023
The returned order is undefined. And currently if an item doesn't
2024
exist, it isn't included in the output.
2028
for file_id in file_ids:
2029
entry = self._fileid_to_entry_cache.get(file_id, None)
2031
remaining.append(file_id)
2033
result.append(entry)
2034
file_keys = [StaticTuple(f,).intern() for f in remaining]
2035
for file_key, value in self.id_to_entry.iteritems(file_keys):
2036
entry = self._bytes_to_entry(value)
2037
result.append(entry)
2038
self._fileid_to_entry_cache[entry.file_id] = entry
2041
def has_id(self, file_id):
2042
# Perhaps have an explicit 'contains' method on CHKMap ?
2043
if self._fileid_to_entry_cache.get(file_id, None) is not None:
2046
self.id_to_entry.iteritems([StaticTuple(file_id,)]))) == 1
2048
def is_root(self, file_id):
2049
return file_id == self.root_id
2051
def _iter_file_id_parents(self, file_id):
2052
"""Yield the parents of file_id up to the root."""
2053
while file_id is not None:
2057
raise errors.NoSuchId(tree=self, file_id=file_id)
2059
file_id = ie.parent_id
2062
"""Iterate over all file-ids."""
2063
for key, _ in self.id_to_entry.iteritems():
2066
def iter_just_entries(self):
2067
"""Iterate over all entries.
2069
Unlike iter_entries(), just the entries are returned (not (path, ie))
2070
and the order of entries is undefined.
2072
XXX: We may not want to merge this into bzr.dev.
2074
for key, entry in self.id_to_entry.iteritems():
2076
ie = self._fileid_to_entry_cache.get(file_id, None)
2078
ie = self._bytes_to_entry(entry)
2079
self._fileid_to_entry_cache[file_id] = ie
2082
def iter_changes(self, basis):
2083
"""Generate a Tree.iter_changes change list between this and basis.
2085
:param basis: Another CHKInventory.
2086
:return: An iterator over the changes between self and basis, as per
2087
tree.iter_changes().
2089
# We want: (file_id, (path_in_source, path_in_target),
2090
# changed_content, versioned, parent, name, kind,
2092
for key, basis_value, self_value in \
2093
self.id_to_entry.iter_changes(basis.id_to_entry):
2095
if basis_value is not None:
2096
basis_entry = basis._bytes_to_entry(basis_value)
2097
path_in_source = basis.id2path(file_id)
2098
basis_parent = basis_entry.parent_id
2099
basis_name = basis_entry.name
2100
basis_executable = basis_entry.executable
2102
path_in_source = None
2105
basis_executable = None
2106
if self_value is not None:
2107
self_entry = self._bytes_to_entry(self_value)
2108
path_in_target = self.id2path(file_id)
2109
self_parent = self_entry.parent_id
2110
self_name = self_entry.name
2111
self_executable = self_entry.executable
2113
path_in_target = None
2116
self_executable = None
2117
if basis_value is None:
2119
kind = (None, self_entry.kind)
2120
versioned = (False, True)
2121
elif self_value is None:
2123
kind = (basis_entry.kind, None)
2124
versioned = (True, False)
2126
kind = (basis_entry.kind, self_entry.kind)
2127
versioned = (True, True)
2128
changed_content = False
2129
if kind[0] != kind[1]:
2130
changed_content = True
2131
elif kind[0] == 'file':
2132
if (self_entry.text_size != basis_entry.text_size or
2133
self_entry.text_sha1 != basis_entry.text_sha1):
2134
changed_content = True
2135
elif kind[0] == 'symlink':
2136
if self_entry.symlink_target != basis_entry.symlink_target:
2137
changed_content = True
2138
elif kind[0] == 'tree-reference':
2139
if (self_entry.reference_revision !=
2140
basis_entry.reference_revision):
2141
changed_content = True
2142
parent = (basis_parent, self_parent)
2143
name = (basis_name, self_name)
2144
executable = (basis_executable, self_executable)
2145
if (not changed_content
2146
and parent[0] == parent[1]
2147
and name[0] == name[1]
2148
and executable[0] == executable[1]):
2149
# Could happen when only the revision changed for a directory
2152
yield (file_id, (path_in_source, path_in_target), changed_content,
2153
versioned, parent, name, kind, executable)
2156
"""Return the number of entries in the inventory."""
2157
return len(self.id_to_entry)
2159
def _make_delta(self, old):
2160
"""Make an inventory delta from two inventories."""
2161
if type(old) != CHKInventory:
2162
return CommonInventory._make_delta(self, old)
2164
for key, old_value, self_value in \
2165
self.id_to_entry.iter_changes(old.id_to_entry):
2167
if old_value is not None:
2168
old_path = old.id2path(file_id)
2171
if self_value is not None:
2172
entry = self._bytes_to_entry(self_value)
2173
self._fileid_to_entry_cache[file_id] = entry
2174
new_path = self.id2path(file_id)
2178
delta.append((old_path, new_path, file_id, entry))
2181
def path2id(self, relpath):
2182
"""See CommonInventory.path2id()."""
2183
# TODO: perhaps support negative hits?
2184
result = self._path_to_fileid_cache.get(relpath, None)
2185
if result is not None:
2187
if isinstance(relpath, basestring):
2188
names = osutils.splitpath(relpath)
2191
current_id = self.root_id
2192
if current_id is None:
2194
parent_id_index = self.parent_id_basename_to_file_id
2196
for basename in names:
2197
if cur_path is None:
2200
cur_path = cur_path + '/' + basename
2201
basename_utf8 = basename.encode('utf8')
2202
file_id = self._path_to_fileid_cache.get(cur_path, None)
2204
key_filter = [StaticTuple(current_id, basename_utf8)]
2205
items = parent_id_index.iteritems(key_filter)
2206
for (parent_id, name_utf8), file_id in items:
2207
if parent_id != current_id or name_utf8 != basename_utf8:
2208
raise errors.BzrError("corrupt inventory lookup! "
2209
"%r %r %r %r" % (parent_id, current_id, name_utf8,
2214
self._path_to_fileid_cache[cur_path] = file_id
2215
current_id = file_id
2219
"""Serialise the inventory to lines."""
2220
lines = ["chkinventory:\n"]
2221
if self._search_key_name != 'plain':
2222
# custom ordering grouping things that don't change together
2223
lines.append('search_key_name: %s\n' % (self._search_key_name,))
2224
lines.append("root_id: %s\n" % self.root_id)
2225
lines.append('parent_id_basename_to_file_id: %s\n' %
2226
(self.parent_id_basename_to_file_id.key()[0],))
2227
lines.append("revision_id: %s\n" % self.revision_id)
2228
lines.append("id_to_entry: %s\n" % (self.id_to_entry.key()[0],))
2230
lines.append("revision_id: %s\n" % self.revision_id)
2231
lines.append("root_id: %s\n" % self.root_id)
2232
if self.parent_id_basename_to_file_id is not None:
2233
lines.append('parent_id_basename_to_file_id: %s\n' %
2234
(self.parent_id_basename_to_file_id.key()[0],))
2235
lines.append("id_to_entry: %s\n" % (self.id_to_entry.key()[0],))
2240
"""Get the root entry."""
2241
return self[self.root_id]
2244
class CHKInventoryDirectory(InventoryDirectory):
2245
"""A directory in an inventory."""
2247
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
2248
'text_id', 'parent_id', '_children', 'executable',
2249
'revision', 'symlink_target', 'reference_revision',
2252
def __init__(self, file_id, name, parent_id, chk_inventory):
2253
# Don't call InventoryDirectory.__init__ - it isn't right for this
2255
InventoryEntry.__init__(self, file_id, name, parent_id)
2256
self._children = None
2257
self.kind = 'directory'
2258
self._chk_inventory = chk_inventory
2262
"""Access the list of children of this directory.
2264
With a parent_id_basename_to_file_id index, loads all the children,
2265
without loads the entire index. Without is bad. A more sophisticated
2266
proxy object might be nice, to allow partial loading of children as
2267
well when specific names are accessed. (So path traversal can be
2268
written in the obvious way but not examine siblings.).
2270
if self._children is not None:
2271
return self._children
2272
# No longer supported
2273
if self._chk_inventory.parent_id_basename_to_file_id is None:
2274
raise AssertionError("Inventories without"
2275
" parent_id_basename_to_file_id are no longer supported")
2277
# XXX: Todo - use proxy objects for the children rather than loading
2278
# all when the attribute is referenced.
2279
parent_id_index = self._chk_inventory.parent_id_basename_to_file_id
2281
for (parent_id, name_utf8), file_id in parent_id_index.iteritems(
2282
key_filter=[StaticTuple(self.file_id,)]):
2283
child_keys.add(StaticTuple(file_id,))
2285
for file_id_key in child_keys:
2286
entry = self._chk_inventory._fileid_to_entry_cache.get(
2287
file_id_key[0], None)
2288
if entry is not None:
2289
result[entry.name] = entry
2290
cached.add(file_id_key)
2291
child_keys.difference_update(cached)
2292
# populate; todo: do by name
2293
id_to_entry = self._chk_inventory.id_to_entry
2294
for file_id_key, bytes in id_to_entry.iteritems(child_keys):
2295
entry = self._chk_inventory._bytes_to_entry(bytes)
2296
result[entry.name] = entry
2297
self._chk_inventory._fileid_to_entry_cache[file_id_key[0]] = entry
2298
self._children = result
2302
'directory': InventoryDirectory,
2303
'file': InventoryFile,
2304
'symlink': InventoryLink,
2305
'tree-reference': TreeReference
2308
def make_entry(kind, name, parent_id, file_id=None):
2309
"""Create an inventory entry.
2311
:param kind: the type of inventory entry to create.
2312
:param name: the basename of the entry.
2313
:param parent_id: the parent_id of the entry.
2314
:param file_id: the file_id to use. if None, one will be created.
2317
file_id = generate_ids.gen_file_id(name)
2318
name = ensure_normalized_name(name)
2320
factory = entry_factory[kind]
2322
raise errors.BadFileKindError(name, kind)
2323
return factory(file_id, name, parent_id)
2326
def ensure_normalized_name(name):
2329
:raises InvalidNormalization: When name is not normalized, and cannot be
2330
accessed on this platform by the normalized path.
2331
:return: The NFC normalised version of name.
2333
#------- This has been copied to bzrlib.dirstate.DirState.add, please
2334
# keep them synchronised.
2335
# we dont import normalized_filename directly because we want to be
2336
# able to change the implementation at runtime for tests.
2337
norm_name, can_access = osutils.normalized_filename(name)
2338
if norm_name != name:
2342
# TODO: jam 20060701 This would probably be more useful
2343
# if the error was raised with the full path
2344
raise errors.InvalidNormalization(name)
1138
2348
_NAME_RE = None
1140
2350
def is_valid_name(name):
1141
2351
global _NAME_RE
1142
if _NAME_RE == None:
2352
if _NAME_RE is None:
1143
2353
_NAME_RE = re.compile(r'^[^/\\]+$')
1145
2355
return bool(_NAME_RE.match(name))
2358
def _check_delta_unique_ids(delta):
2359
"""Decorate a delta and check that the file ids in it are unique.
2361
:return: A generator over delta.
2365
length = len(ids) + 1
2367
if len(ids) != length:
2368
raise errors.InconsistentDelta(item[0] or item[1], item[2],
2373
def _check_delta_unique_new_paths(delta):
2374
"""Decorate a delta and check that the new paths in it are unique.
2376
:return: A generator over delta.
2380
length = len(paths) + 1
2382
if path is not None:
2384
if len(paths) != length:
2385
raise errors.InconsistentDelta(path, item[2], "repeated path")
2389
def _check_delta_unique_old_paths(delta):
2390
"""Decorate a delta and check that the old paths in it are unique.
2392
:return: A generator over delta.
2396
length = len(paths) + 1
2398
if path is not None:
2400
if len(paths) != length:
2401
raise errors.InconsistentDelta(path, item[2], "repeated path")
2405
def _check_delta_ids_are_valid(delta):
2406
"""Decorate a delta and check that the ids in it are valid.
2408
:return: A generator over delta.
2413
raise errors.InconsistentDelta(item[0] or item[1], item[2],
2414
"entry with file_id None %r" % entry)
2415
if type(item[2]) != str:
2416
raise errors.InconsistentDelta(item[0] or item[1], item[2],
2417
"entry with non bytes file_id %r" % entry)
2421
def _check_delta_ids_match_entry(delta):
2422
"""Decorate a delta and check that the ids in it match the entry.file_id.
2424
:return: A generator over delta.
2428
if entry is not None:
2429
if entry.file_id != item[2]:
2430
raise errors.InconsistentDelta(item[0] or item[1], item[2],
2431
"mismatched id with %r" % entry)
2435
def _check_delta_new_path_entry_both_or_None(delta):
2436
"""Decorate a delta and check that the new_path and entry are paired.
2438
:return: A generator over delta.
2443
if new_path is None and entry is not None:
2444
raise errors.InconsistentDelta(item[0], item[1],
2445
"Entry with no new_path")
2446
if new_path is not None and entry is None:
2447
raise errors.InconsistentDelta(new_path, item[1],
2448
"new_path with no entry")
added
removed
bzrlib/inventory.py
