711
713
return compatible
714
class Inventory(object):
716
class CommonInventory(object):
717
"""Basic inventory logic, defined in terms of primitives like has_id."""
719
def __contains__(self, file_id):
720
"""True if this entry contains a file with given id.
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)
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
734
return self.has_id(file_id)
736
def id2path(self, file_id):
737
"""Return as a string the path to file_id.
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')
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]))
750
def iter_entries(self, from_dir=None):
751
"""Return (path, entry) pairs, in order by name."""
753
if self.root is None:
757
elif isinstance(from_dir, basestring):
758
from_dir = self[from_dir]
760
# unrolling the recursive called changed the time from
761
# 440ms/663ms (inline/total) to 116ms/116ms
762
children = from_dir.children.items()
764
children = collections.deque(children)
765
stack = [(u'', children)]
767
from_dir_relpath, children = stack[-1]
770
name, ie = children.popleft()
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
776
path = from_dir_relpath + '/' + name
780
if ie.kind != 'directory':
783
# But do this child first
784
new_children = ie.children.items()
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
791
# if we finished all children, pop it off the stack
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.
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.
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
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?
813
if self.root is None:
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]
820
yield self.id2path(file_id), self[file_id]
823
if (specific_file_ids is None or yield_parents or
824
self.root.file_id in specific_file_ids):
826
elif isinstance(from_dir, basestring):
827
from_dir = self[from_dir]
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.
834
def add_ancestors(file_id):
835
if file_id not in byid:
837
parent_id = byid[file_id].parent_id
838
if parent_id is None:
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)
848
stack = [(u'', from_dir)]
850
cur_relpath, cur_dir = stack.pop()
853
for child_name, child_ie in sorted(cur_dir.children.iteritems()):
855
child_relpath = cur_relpath + child_name
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
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))
867
def _make_delta(self, old):
868
"""Make an inventory delta from two inventories."""
871
adds = new_ids - old_ids
872
deletes = old_ids - new_ids
873
common = old_ids.intersection(new_ids)
875
for file_id in deletes:
876
delta.append((old.id2path(file_id), None, file_id, None))
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]))
885
def _get_mutable_inventory(self):
886
"""Returns a mutable copy of the object.
888
Some inventories are immutable, yet working trees, for example, needs
889
to mutate exisiting inventories instead of creating a new one.
891
raise NotImplementedError(self._get_mutable_inventory)
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)
898
"""Return list of (path, ie) for all entries except the root.
900
This may be faster than iter_entries.
903
def descend(dir_ie, dir_path):
904
kids = dir_ie.children.items()
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)
912
descend(self.root, u'')
915
def directories(self):
916
"""Return (path, entry) pairs for all directories, including the root.
919
def descend(parent_ie, parent_path):
920
accum.append((parent_path, parent_ie))
922
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
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'')
931
def path2id(self, name):
932
"""Walk down through directories to return entry of last component.
934
names may be either a list of path components, or a single
935
string, in which case it is automatically split.
937
This returns the entry of the last component in the path,
938
which may be either a file or a directory.
940
Returns None IFF the path is not found.
942
if isinstance(name, basestring):
943
name = osutils.splitpath(name)
945
# mutter("lookup path %r" % name)
949
except errors.NoSuchId:
950
# root doesn't exist yet so nothing else can
956
children = getattr(parent, 'children', None)
965
return parent.file_id
967
def filter(self, specific_fileids):
968
"""Get an inventory view filtered against a set of file-ids.
970
Children of directories and parents are included.
972
The result may or may not reference the underlying inventory
973
so it should be treated as immutable.
975
interesting_parents = set()
976
for fileid in specific_fileids:
978
interesting_parents.update(self.get_idpath(fileid))
979
except errors.NoSuchId:
980
# This fileid is not in the inventory - that's ok
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:
997
other.add(entry.copy())
1000
def get_idpath(self, file_id):
1001
"""Return a list of file_ids for the path to an entry.
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.
1009
for parent in self._iter_file_id_parents(file_id):
1010
p.insert(0, parent.file_id)
1014
class Inventory(CommonInventory):
715
1015
"""Inventory of versioned files in a tree.
717
1017
This describes which file_id is present at each point in the tree,
880
1184
def _get_mutable_inventory(self):
881
"""Returns a mutable copy of the object.
883
Some inventories are immutable, yet working trees, for example, needs
884
to mutate exisiting inventories instead of creating a new one.
1185
"""See CommonInventory._get_mutable_inventory."""
886
1186
return deepcopy(self)
888
1188
def __iter__(self):
1189
"""Iterate over all file-ids."""
889
1190
return iter(self._byid)
1192
def iter_just_entries(self):
1193
"""Iterate over all entries.
1195
Unlike iter_entries(), just the entries are returned (not (path, ie))
1196
and the order of entries is undefined.
1198
XXX: We may not want to merge this into bzr.dev.
1200
if self.root is None:
1202
for _, ie in self._byid.iteritems():
891
1205
def __len__(self):
892
1206
"""Returns number of entries."""
893
1207
return len(self._byid)
895
def iter_entries(self, from_dir=None):
896
"""Return (path, entry) pairs, in order by name."""
898
if self.root is None:
902
elif isinstance(from_dir, basestring):
903
from_dir = self._byid[from_dir]
905
# unrolling the recursive called changed the time from
906
# 440ms/663ms (inline/total) to 116ms/116ms
907
children = from_dir.children.items()
909
children = collections.deque(children)
910
stack = [(u'', children)]
912
from_dir_relpath, children = stack[-1]
915
name, ie = children.popleft()
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
921
path = from_dir_relpath + '/' + name
925
if ie.kind != 'directory':
928
# But do this child first
929
new_children = ie.children.items()
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
936
# if we finished all children, pop it off the stack
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.
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.
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
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?
958
if self.root is None:
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]
965
yield self.id2path(file_id), self[file_id]
968
if (specific_file_ids is None or yield_parents or
969
self.root.file_id in specific_file_ids):
971
elif isinstance(from_dir, basestring):
972
from_dir = self._byid[from_dir]
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.
979
def add_ancestors(file_id):
980
if file_id not in byid:
982
parent_id = byid[file_id].parent_id
983
if parent_id is None:
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)
993
stack = [(u'', from_dir)]
995
cur_relpath, cur_dir = stack.pop()
998
for child_name, child_ie in sorted(cur_dir.children.iteritems()):
1000
child_relpath = cur_relpath + child_name
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
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))
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)
1017
"""Return list of (path, ie) for all entries except the root.
1019
This may be faster than iter_entries.
1022
def descend(dir_ie, dir_path):
1023
kids = dir_ie.children.items()
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)
1031
descend(self.root, u'')
1034
def directories(self):
1035
"""Return (path, entry) pairs for all directories, including the root.
1038
def descend(parent_ie, parent_path):
1039
accum.append((parent_path, parent_ie))
1041
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
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'')
1050
def __contains__(self, file_id):
1051
"""True if this entry contains a file with given id.
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)
1061
return (file_id in self._byid)
1063
1209
def __getitem__(self, file_id):
1064
1210
"""Return the entry for given file_id.
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
1346
1433
def is_root(self, file_id):
1347
1434
return self.root is not None and file_id == self.root.file_id
1349
def filter(self, specific_fileids):
1350
"""Get an inventory view filtered against a set of file-ids.
1352
Children of directories and parents are included.
1354
The result may or may not reference the underlying inventory
1355
so it should be treated as immutable.
1357
interesting_parents = set()
1358
for fileid in specific_fileids:
1437
class CHKInventory(CommonInventory):
1438
"""An inventory persisted in a CHK store.
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.
1445
Internally, a CHKInventory has one or two CHKMaps:
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)
1451
The second map is optional and not present in early CHkRepository's.
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
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
1464
def _entry_to_bytes(self, entry):
1465
"""Serialise entry as a single bytestring.
1467
:param Entry: An inventory entry.
1468
:return: A bytestring for the entry.
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
1479
if entry.parent_id is not None:
1480
parent_str = entry.parent_id
1483
name_str = entry.name.encode("utf8")
1484
if entry.kind == 'file':
1485
if entry.executable:
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)
1504
raise ValueError("unknown kind %r" % entry.kind)
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])
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'),
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'),
1529
elif sections[0].startswith("symlink: "):
1530
result = InventoryLink(sections[0][9:],
1531
sections[2].decode('utf8'),
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'),
1538
result.reference_revision = sections[4]
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
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:
1552
inv = Inventory(self.root_id, self.revision_id)
1553
for path, inv_entry in entries:
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.
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.
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
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:
1601
result.root_id = file_id
1602
if new_path is None:
1607
if propagate_caches:
1609
del result._path_to_fileid_cache[old_path]
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:
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:
1628
old_entry = self[file_id]
1629
old_key = self._parent_id_basename_key(old_entry)
1630
if new_path is None:
1634
new_key = self._parent_id_basename_key(entry)
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)
1646
def deserialise(klass, chk_store, bytes, expected_revision_id):
1647
"""Deserialise a CHKInventory.
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
1653
:return: A CHKInventory
1655
lines = bytes.split('\n')
1657
raise AssertionError('bytes to deserialize must end with an eol')
1659
if lines[0] != 'chkinventory:':
1660
raise ValueError("not a serialised CHKInventory: %r" % bytes)
1662
allowed_keys = frozenset(['root_id', 'revision_id', 'search_key_name',
1663
'parent_id_basename_to_file_id',
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'
1671
raise errors.BzrError('Duplicate key in inventory: %r\n%r'
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']
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)
1691
result.parent_id_basename_to_file_id = None
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))
1701
def from_inventory(klass, chk_store, inventory, maximum_size=0, search_key_name='plain'):
1702
"""Create a CHKInventory from an existing inventory.
1704
The content of inventory is copied into the chk_store, and a
1705
CHKInventory referencing that is returned.
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
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)
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(
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),
1731
result.id_to_entry.apply_delta(file_id_delta)
1732
result.parent_id_basename_to_file_id.apply_delta(parent_id_delta)
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
1741
return parent_id, entry.name.encode('utf8')
1743
def __getitem__(self, file_id):
1744
"""map a single file_id -> InventoryEntry."""
1746
raise errors.NoSuchId(self, file_id)
1747
result = self._fileid_to_entry_cache.get(file_id, None)
1748
if result is not None:
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)
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:
1761
return len(list(self.id_to_entry.iteritems([(file_id,)]))) == 1
1763
def is_root(self, file_id):
1764
return file_id == self.root_id
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:
1360
interesting_parents.update(self.get_idpath(fileid))
1361
except errors.NoSuchId:
1362
# This fileid is not in the inventory - that's ok
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:
1772
raise errors.NoSuchId(tree=self, file_id=file_id)
1774
file_id = ie.parent_id
1777
"""Iterate over all file-ids."""
1778
for key, _ in self.id_to_entry.iteritems():
1781
def iter_just_entries(self):
1782
"""Iterate over all entries.
1784
Unlike iter_entries(), just the entries are returned (not (path, ie))
1785
and the order of entries is undefined.
1787
XXX: We may not want to merge this into bzr.dev.
1789
for key, entry in self.id_to_entry.iteritems():
1791
ie = self._fileid_to_entry_cache.get(file_id, None)
1793
ie = self._bytes_to_entry(entry)
1794
self._fileid_to_entry_cache[file_id] = ie
1797
def iter_changes(self, basis):
1798
"""Generate a Tree.iter_changes change list between this and basis.
1800
:param basis: Another CHKInventory.
1801
:return: An iterator over the changes between self and basis, as per
1802
tree.iter_changes().
1804
# We want: (file_id, (path_in_source, path_in_target),
1805
# changed_content, versioned, parent, name, kind,
1807
for key, basis_value, self_value in \
1808
self.id_to_entry.iter_changes(basis.id_to_entry):
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
1817
path_in_source = 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
1828
path_in_target = None
1831
self_executable = None
1832
if basis_value is None:
1834
kind = (None, self_entry.kind)
1835
versioned = (False, True)
1836
elif self_value is None:
1838
kind = (basis_entry.kind, None)
1839
versioned = (True, False)
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
1379
other.add(entry.copy())
1867
yield (file_id, (path_in_source, path_in_target), changed_content,
1868
versioned, parent, name, kind, executable)
1871
"""Return the number of entries in the inventory."""
1872
return len(self.id_to_entry)
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)
1879
for key, old_value, self_value in \
1880
self.id_to_entry.iter_changes(old.id_to_entry):
1882
if old_value is not None:
1883
old_path = old.id2path(file_id)
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)
1893
delta.append((old_path, new_path, file_id, entry))
1896
def path2id(self, name):
1897
"""See CommonInventory.path2id()."""
1898
result = self._path_to_fileid_cache.get(name, None)
1900
result = CommonInventory.path2id(self, name)
1901
self._path_to_fileid_cache[name] = result
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())
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())
1926
"""Get the root entry."""
1927
return self[self.root_id]
1930
class CHKInventoryDirectory(InventoryDirectory):
1931
"""A directory in an inventory."""
1933
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
1934
'text_id', 'parent_id', '_children', 'executable',
1935
'revision', 'symlink_target', 'reference_revision',
1938
def __init__(self, file_id, name, parent_id, chk_inventory):
1939
# Don't call InventoryDirectory.__init__ - it isn't right for this
1941
InventoryEntry.__init__(self, file_id, name, parent_id)
1942
self._children = None
1943
self.kind = 'directory'
1944
self._chk_inventory = chk_inventory
1948
"""Access the list of children of this directory.
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.).
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")
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
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,))
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
1383
1987
entry_factory = {
1384
1988
'directory': InventoryDirectory,