713
816
return compatible
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]))
819
class Inventory(object):
820
"""Inventory of versioned files in a tree.
822
This describes which file_id is present at each point in the tree,
823
and possibly the SHA-1 or other information about the file.
824
Entries can be looked up either by path or by file_id.
826
The inventory represents a typical unix file tree, with
827
directories containing files and subdirectories. We never store
828
the full path to a file, because renaming a directory implicitly
829
moves all of its contents. This class internally maintains a
830
lookup tree that allows the children under a directory to be
833
InventoryEntry objects must not be modified after they are
834
inserted, other than through the Inventory API.
836
>>> inv = Inventory()
837
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
838
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
839
>>> inv['123-123'].name
842
May be treated as an iterator or set to look up file ids:
844
>>> bool(inv.path2id('hello.c'))
849
May also look up by name:
851
>>> [x[0] for x in inv.iter_entries()]
853
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
854
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
855
Traceback (most recent call last):
856
BzrError: parent_id {TREE_ROOT} not in inventory
857
>>> inv.add(InventoryFile('123-123', 'hello.c', 'TREE_ROOT-12345678-12345678'))
858
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT-12345678-12345678', sha1=None, len=None)
860
def __init__(self, root_id=ROOT_ID, revision_id=None):
861
"""Create or read an inventory.
863
If a working directory is specified, the inventory is read
864
from there. If the file is specified, read from that. If not,
865
the inventory is created empty.
867
The inventory is created with a default root directory, with
870
if root_id is not None:
871
assert root_id.__class__ == str
872
self._set_root(InventoryDirectory(root_id, u'', None))
876
self.revision_id = revision_id
879
return "<Inventory object at %x, contents=%r>" % (id(self), self._byid)
881
def apply_delta(self, delta):
882
"""Apply a delta to this inventory.
884
:param delta: A list of changes to apply. After all the changes are
885
applied the final inventory must be internally consistent, but it
886
is ok to supply changes which, if only half-applied would have an
887
invalid result - such as supplying two changes which rename two
888
files, 'A' and 'B' with each other : [('A', 'B', 'A-id', a_entry),
889
('B', 'A', 'B-id', b_entry)].
891
Each change is a tuple, of the form (old_path, new_path, file_id,
894
When new_path is None, the change indicates the removal of an entry
895
from the inventory and new_entry will be ignored (using None is
896
appropriate). If new_path is not None, then new_entry must be an
897
InventoryEntry instance, which will be incorporated into the
898
inventory (and replace any existing entry with the same file id).
900
When old_path is None, the change indicates the addition of
901
a new entry to the inventory.
903
When neither new_path nor old_path are None, the change is a
904
modification to an entry, such as a rename, reparent, kind change
907
The children attribute of new_entry is ignored. This is because
908
this method preserves children automatically across alterations to
909
the parent of the children, and cases where the parent id of a
910
child is changing require the child to be passed in as a separate
911
change regardless. E.g. in the recursive deletion of a directory -
912
the directory's children must be included in the delta, or the
913
final inventory will be invalid.
916
# Remove all affected items which were in the original inventory,
917
# starting with the longest paths, thus ensuring parents are examined
918
# after their children, which means that everything we examine has no
919
# modified children remaining by the time we examine it.
920
for old_path, file_id in sorted(((op, f) for op, np, f, e in delta
921
if op is not None), reverse=True):
922
if file_id not in self:
925
# Preserve unaltered children of file_id for later reinsertion.
926
children[file_id] = getattr(self[file_id], 'children', {})
927
# Remove file_id and the unaltered children. If file_id is not
928
# being deleted it will be reinserted back later.
929
self.remove_recursive_id(file_id)
930
# Insert all affected which should be in the new inventory, reattaching
931
# their children if they had any. This is done from shortest path to
932
# longest, ensuring that items which were modified and whose parents in
933
# the resulting inventory were also modified, are inserted after their
935
for new_path, new_entry in sorted((np, e) for op, np, f, e in
936
delta if np is not None):
937
if new_entry.kind == 'directory':
938
new_entry.children = children.get(new_entry.file_id, {})
941
def _set_root(self, ie):
943
self._byid = {self.root.file_id: self.root}
946
# TODO: jam 20051218 Should copy also copy the revision_id?
947
entries = self.iter_entries()
948
if self.root is None:
949
return Inventory(root_id=None)
950
other = Inventory(entries.next()[1].file_id)
951
# copy recursively so we know directories will be added before
952
# their children. There are more efficient ways than this...
953
for path, entry in entries:
954
other.add(entry.copy())
958
return iter(self._byid)
961
"""Returns number of entries."""
962
return len(self._byid)
750
964
def iter_entries(self, from_dir=None):
751
965
"""Return (path, entry) pairs, in order by name."""
927
1115
descend(child_ie, child_path)
928
1116
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):
1015
"""Inventory of versioned files in a tree.
1017
This describes which file_id is present at each point in the tree,
1018
and possibly the SHA-1 or other information about the file.
1019
Entries can be looked up either by path or by file_id.
1021
The inventory represents a typical unix file tree, with
1022
directories containing files and subdirectories. We never store
1023
the full path to a file, because renaming a directory implicitly
1024
moves all of its contents. This class internally maintains a
1025
lookup tree that allows the children under a directory to be
1028
InventoryEntry objects must not be modified after they are
1029
inserted, other than through the Inventory API.
1031
>>> inv = Inventory()
1032
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
1033
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
1034
>>> inv['123-123'].name
1037
May be treated as an iterator or set to look up file ids:
1039
>>> bool(inv.path2id('hello.c'))
1041
>>> '123-123' in inv
1044
May also look up by name:
1046
>>> [x[0] for x in inv.iter_entries()]
1048
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
1049
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
1050
Traceback (most recent call last):
1051
BzrError: parent_id {TREE_ROOT} not in inventory
1052
>>> inv.add(InventoryFile('123-123', 'hello.c', 'TREE_ROOT-12345678-12345678'))
1053
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT-12345678-12345678', sha1=None, len=None, revision=None)
1055
def __init__(self, root_id=ROOT_ID, revision_id=None):
1056
"""Create or read an inventory.
1058
If a working directory is specified, the inventory is read
1059
from there. If the file is specified, read from that. If not,
1060
the inventory is created empty.
1062
The inventory is created with a default root directory, with
1065
if root_id is not None:
1066
self._set_root(InventoryDirectory(root_id, u'', None))
1070
self.revision_id = revision_id
1073
# More than one page of ouput is not useful anymore to debug
1076
contents = repr(self._byid)
1077
if len(contents) > max_len:
1078
contents = contents[:(max_len-len(closing))] + closing
1079
return "<Inventory object at %x, contents=%r>" % (id(self), contents)
1081
def apply_delta(self, delta):
1082
"""Apply a delta to this inventory.
1084
:param delta: A list of changes to apply. After all the changes are
1085
applied the final inventory must be internally consistent, but it
1086
is ok to supply changes which, if only half-applied would have an
1087
invalid result - such as supplying two changes which rename two
1088
files, 'A' and 'B' with each other : [('A', 'B', 'A-id', a_entry),
1089
('B', 'A', 'B-id', b_entry)].
1091
Each change is a tuple, of the form (old_path, new_path, file_id,
1094
When new_path is None, the change indicates the removal of an entry
1095
from the inventory and new_entry will be ignored (using None is
1096
appropriate). If new_path is not None, then new_entry must be an
1097
InventoryEntry instance, which will be incorporated into the
1098
inventory (and replace any existing entry with the same file id).
1100
When old_path is None, the change indicates the addition of
1101
a new entry to the inventory.
1103
When neither new_path nor old_path are None, the change is a
1104
modification to an entry, such as a rename, reparent, kind change
1107
The children attribute of new_entry is ignored. This is because
1108
this method preserves children automatically across alterations to
1109
the parent of the children, and cases where the parent id of a
1110
child is changing require the child to be passed in as a separate
1111
change regardless. E.g. in the recursive deletion of a directory -
1112
the directory's children must be included in the delta, or the
1113
final inventory will be invalid.
1115
Note that a file_id must only appear once within a given delta.
1116
An AssertionError is raised otherwise.
1118
# Check that the delta is legal. It would be nice if this could be
1119
# done within the loops below but it's safer to validate the delta
1120
# before starting to mutate the inventory.
1121
unique_file_ids = set([f for _, _, f, _ in delta])
1122
if len(unique_file_ids) != len(delta):
1123
raise AssertionError("a file-id appears multiple times in %r"
1128
# Remove all affected items which were in the original inventory,
1129
# starting with the longest paths, thus ensuring parents are examined
1130
# after their children, which means that everything we examine has no
1131
# modified children remaining by the time we examine it.
1132
for old_path, file_id in sorted(((op, f) for op, np, f, e in delta
1133
if op is not None), reverse=True):
1134
if file_id not in self:
1137
# Preserve unaltered children of file_id for later reinsertion.
1138
file_id_children = getattr(self[file_id], 'children', {})
1139
if len(file_id_children):
1140
children[file_id] = file_id_children
1141
# Remove file_id and the unaltered children. If file_id is not
1142
# being deleted it will be reinserted back later.
1143
self.remove_recursive_id(file_id)
1144
# Insert all affected which should be in the new inventory, reattaching
1145
# their children if they had any. This is done from shortest path to
1146
# longest, ensuring that items which were modified and whose parents in
1147
# the resulting inventory were also modified, are inserted after their
1149
for new_path, new_entry in sorted((np, e) for op, np, f, e in
1150
delta if np is not None):
1151
if new_entry.kind == 'directory':
1152
# Pop the child which to allow detection of children whose
1153
# parents were deleted and which were not reattached to a new
1155
replacement = InventoryDirectory(new_entry.file_id,
1156
new_entry.name, new_entry.parent_id)
1157
replacement.revision = new_entry.revision
1158
replacement.children = children.pop(replacement.file_id, {})
1159
new_entry = replacement
1162
# Get the parent id that was deleted
1163
parent_id, children = children.popitem()
1164
raise errors.InconsistentDelta("<deleted>", parent_id,
1165
"The file id was deleted but its children were not deleted.")
1167
def _set_root(self, ie):
1169
self._byid = {self.root.file_id: self.root}
1172
# TODO: jam 20051218 Should copy also copy the revision_id?
1173
entries = self.iter_entries()
1174
if self.root is None:
1175
return Inventory(root_id=None)
1176
other = Inventory(entries.next()[1].file_id)
1177
other.root.revision = self.root.revision
1178
# copy recursively so we know directories will be added before
1179
# their children. There are more efficient ways than this...
1180
for path, entry in entries:
1181
other.add(entry.copy())
1184
def _get_mutable_inventory(self):
1185
"""See CommonInventory._get_mutable_inventory."""
1186
return deepcopy(self)
1189
"""Iterate over all file-ids."""
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.
1119
def __contains__(self, file_id):
1120
"""True if this entry contains a file with given id.
1198
XXX: We may not want to merge this into bzr.dev.
1122
>>> inv = Inventory()
1123
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1124
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
1200
if self.root is None:
1202
for _, ie in self._byid.iteritems():
1206
"""Returns number of entries."""
1207
return len(self._byid)
1130
return (file_id in self._byid)
1209
1132
def __getitem__(self, file_id):
1210
1133
"""Return the entry for given file_id.
1212
1135
>>> inv = Inventory()
1213
1136
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
1214
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
1137
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
1215
1138
>>> inv['123123'].name
1434
1392
return self.root is not None and file_id == self.root.file_id
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
inv = Inventory(None, self.revision_id)
1551
for path, inv_entry in entries:
1552
inv.add(inv_entry.copy())
1555
def create_by_apply_delta(self, inventory_delta, new_revision_id,
1556
propagate_caches=False):
1557
"""Create a new CHKInventory by applying inventory_delta to this one.
1559
:param inventory_delta: The inventory delta to apply. See
1560
Inventory.apply_delta for details.
1561
:param new_revision_id: The revision id of the resulting CHKInventory.
1562
:param propagate_caches: If True, the caches for this inventory are
1563
copied to and updated for the result.
1564
:return: The new CHKInventory.
1566
result = CHKInventory(self._search_key_name)
1567
if propagate_caches:
1568
# Just propagate the path-to-fileid cache for now
1569
result._path_to_fileid_cache = dict(self._path_to_fileid_cache.iteritems())
1570
search_key_func = chk_map.search_key_registry.get(self._search_key_name)
1571
self.id_to_entry._ensure_root()
1572
maximum_size = self.id_to_entry._root_node.maximum_size
1573
result.revision_id = new_revision_id
1574
result.id_to_entry = chk_map.CHKMap(
1575
self.id_to_entry._store,
1576
self.id_to_entry.key(),
1577
search_key_func=search_key_func)
1578
result.id_to_entry._ensure_root()
1579
result.id_to_entry._root_node.set_maximum_size(maximum_size)
1580
parent_id_basename_delta = []
1581
if self.parent_id_basename_to_file_id is not None:
1582
result.parent_id_basename_to_file_id = chk_map.CHKMap(
1583
self.parent_id_basename_to_file_id._store,
1584
self.parent_id_basename_to_file_id.key(),
1585
search_key_func=search_key_func)
1586
result.parent_id_basename_to_file_id._ensure_root()
1587
self.parent_id_basename_to_file_id._ensure_root()
1588
result_p_id_root = result.parent_id_basename_to_file_id._root_node
1589
p_id_root = self.parent_id_basename_to_file_id._root_node
1590
result_p_id_root.set_maximum_size(p_id_root.maximum_size)
1591
result_p_id_root._key_width = p_id_root._key_width
1593
result.parent_id_basename_to_file_id = None
1594
result.root_id = self.root_id
1595
id_to_entry_delta = []
1596
for old_path, new_path, file_id, entry in inventory_delta:
1599
result.root_id = file_id
1600
if new_path is None:
1605
if propagate_caches:
1607
del result._path_to_fileid_cache[old_path]
1611
new_key = (file_id,)
1612
new_value = result._entry_to_bytes(entry)
1613
# Update caches. It's worth doing this whether
1614
# we're propagating the old caches or not.
1615
result._path_to_fileid_cache[new_path] = file_id
1616
if old_path is None:
1619
old_key = (file_id,)
1620
id_to_entry_delta.append((old_key, new_key, new_value))
1621
if result.parent_id_basename_to_file_id is not None:
1622
# parent_id, basename changes
1623
if old_path is None:
1626
old_entry = self[file_id]
1627
old_key = self._parent_id_basename_key(old_entry)
1628
if new_path is None:
1632
new_key = self._parent_id_basename_key(entry)
1634
if old_key != new_key:
1635
# If the two keys are the same, the value will be unchanged
1636
# as its always the file id.
1637
parent_id_basename_delta.append((old_key, new_key, new_value))
1638
result.id_to_entry.apply_delta(id_to_entry_delta)
1639
if parent_id_basename_delta:
1640
result.parent_id_basename_to_file_id.apply_delta(parent_id_basename_delta)
1644
def deserialise(klass, chk_store, bytes, expected_revision_id):
1645
"""Deserialise a CHKInventory.
1647
:param chk_store: A CHK capable VersionedFiles instance.
1648
:param bytes: The serialised bytes.
1649
:param expected_revision_id: The revision ID we think this inventory is
1651
:return: A CHKInventory
1653
lines = bytes.split('\n')
1655
raise AssertionError('bytes to deserialize must end with an eol')
1657
if lines[0] != 'chkinventory:':
1658
raise ValueError("not a serialised CHKInventory: %r" % bytes)
1660
allowed_keys = frozenset(['root_id', 'revision_id', 'search_key_name',
1661
'parent_id_basename_to_file_id',
1663
for line in lines[1:]:
1664
key, value = line.split(': ', 1)
1665
if key not in allowed_keys:
1666
raise errors.BzrError('Unknown key in inventory: %r\n%r'
1669
raise errors.BzrError('Duplicate key in inventory: %r\n%r'
1672
revision_id = info['revision_id']
1673
root_id = info['root_id']
1674
search_key_name = info.get('search_key_name', 'plain')
1675
parent_id_basename_to_file_id = info.get(
1676
'parent_id_basename_to_file_id', None)
1677
id_to_entry = info['id_to_entry']
1679
result = CHKInventory(search_key_name)
1680
result.revision_id = revision_id
1681
result.root_id = root_id
1682
search_key_func = chk_map.search_key_registry.get(
1683
result._search_key_name)
1684
if parent_id_basename_to_file_id is not None:
1685
result.parent_id_basename_to_file_id = chk_map.CHKMap(
1686
chk_store, (parent_id_basename_to_file_id,),
1687
search_key_func=search_key_func)
1689
result.parent_id_basename_to_file_id = None
1691
result.id_to_entry = chk_map.CHKMap(chk_store, (id_to_entry,),
1692
search_key_func=search_key_func)
1693
if (result.revision_id,) != expected_revision_id:
1694
raise ValueError("Mismatched revision id and expected: %r, %r" %
1695
(result.revision_id, expected_revision_id))
1699
def from_inventory(klass, chk_store, inventory, maximum_size=0, search_key_name='plain'):
1700
"""Create a CHKInventory from an existing inventory.
1702
The content of inventory is copied into the chk_store, and a
1703
CHKInventory referencing that is returned.
1705
:param chk_store: A CHK capable VersionedFiles instance.
1706
:param inventory: The inventory to copy.
1707
:param maximum_size: The CHKMap node size limit.
1708
:param search_key_name: The identifier for the search key function
1710
result = CHKInventory(search_key_name)
1711
result.revision_id = inventory.revision_id
1712
result.root_id = inventory.root.file_id
1713
search_key_func = chk_map.search_key_registry.get(search_key_name)
1714
result.id_to_entry = chk_map.CHKMap(chk_store, None, search_key_func)
1715
result.id_to_entry._root_node.set_maximum_size(maximum_size)
1717
result.parent_id_basename_to_file_id = chk_map.CHKMap(chk_store,
1718
None, search_key_func)
1719
result.parent_id_basename_to_file_id._root_node.set_maximum_size(
1721
result.parent_id_basename_to_file_id._root_node._key_width = 2
1722
parent_id_delta = []
1723
for path, entry in inventory.iter_entries():
1724
file_id_delta.append((None, (entry.file_id,),
1725
result._entry_to_bytes(entry)))
1726
parent_id_delta.append(
1727
(None, result._parent_id_basename_key(entry),
1729
result.id_to_entry.apply_delta(file_id_delta)
1730
result.parent_id_basename_to_file_id.apply_delta(parent_id_delta)
1733
def _parent_id_basename_key(self, entry):
1734
"""Create a key for a entry in a parent_id_basename_to_file_id index."""
1735
if entry.parent_id is not None:
1736
parent_id = entry.parent_id
1739
return parent_id, entry.name.encode('utf8')
1741
def __getitem__(self, file_id):
1742
"""map a single file_id -> InventoryEntry."""
1744
raise errors.NoSuchId(self, file_id)
1745
result = self._fileid_to_entry_cache.get(file_id, None)
1746
if result is not None:
1749
return self._bytes_to_entry(
1750
self.id_to_entry.iteritems([(file_id,)]).next()[1])
1751
except StopIteration:
1752
# really we're passing an inventory, not a tree...
1753
raise errors.NoSuchId(self, file_id)
1755
def has_id(self, file_id):
1756
# Perhaps have an explicit 'contains' method on CHKMap ?
1757
if self._fileid_to_entry_cache.get(file_id, None) is not None:
1759
return len(list(self.id_to_entry.iteritems([(file_id,)]))) == 1
1761
def is_root(self, file_id):
1762
return file_id == self.root_id
1764
def _iter_file_id_parents(self, file_id):
1765
"""Yield the parents of file_id up to the root."""
1766
while file_id is not None:
1770
raise errors.NoSuchId(tree=self, file_id=file_id)
1772
file_id = ie.parent_id
1775
"""Iterate over all file-ids."""
1776
for key, _ in self.id_to_entry.iteritems():
1779
def iter_just_entries(self):
1780
"""Iterate over all entries.
1782
Unlike iter_entries(), just the entries are returned (not (path, ie))
1783
and the order of entries is undefined.
1785
XXX: We may not want to merge this into bzr.dev.
1787
for key, entry in self.id_to_entry.iteritems():
1789
ie = self._fileid_to_entry_cache.get(file_id, None)
1791
ie = self._bytes_to_entry(entry)
1792
self._fileid_to_entry_cache[file_id] = ie
1795
def iter_changes(self, basis):
1796
"""Generate a Tree.iter_changes change list between this and basis.
1798
:param basis: Another CHKInventory.
1799
:return: An iterator over the changes between self and basis, as per
1800
tree.iter_changes().
1802
# We want: (file_id, (path_in_source, path_in_target),
1803
# changed_content, versioned, parent, name, kind,
1805
for key, basis_value, self_value in \
1806
self.id_to_entry.iter_changes(basis.id_to_entry):
1808
if basis_value is not None:
1809
basis_entry = basis._bytes_to_entry(basis_value)
1810
path_in_source = basis.id2path(file_id)
1811
basis_parent = basis_entry.parent_id
1812
basis_name = basis_entry.name
1813
basis_executable = basis_entry.executable
1815
path_in_source = None
1818
basis_executable = None
1819
if self_value is not None:
1820
self_entry = self._bytes_to_entry(self_value)
1821
path_in_target = self.id2path(file_id)
1822
self_parent = self_entry.parent_id
1823
self_name = self_entry.name
1824
self_executable = self_entry.executable
1826
path_in_target = None
1829
self_executable = None
1830
if basis_value is None:
1832
kind = (None, self_entry.kind)
1833
versioned = (False, True)
1834
elif self_value is None:
1836
kind = (basis_entry.kind, None)
1837
versioned = (True, False)
1839
kind = (basis_entry.kind, self_entry.kind)
1840
versioned = (True, True)
1841
changed_content = False
1842
if kind[0] != kind[1]:
1843
changed_content = True
1844
elif kind[0] == 'file':
1845
if (self_entry.text_size != basis_entry.text_size or
1846
self_entry.text_sha1 != basis_entry.text_sha1):
1847
changed_content = True
1848
elif kind[0] == 'symlink':
1849
if self_entry.symlink_target != basis_entry.symlink_target:
1850
changed_content = True
1851
elif kind[0] == 'tree-reference':
1852
if (self_entry.reference_revision !=
1853
basis_entry.reference_revision):
1854
changed_content = True
1855
parent = (basis_parent, self_parent)
1856
name = (basis_name, self_name)
1857
executable = (basis_executable, self_executable)
1858
if (not changed_content
1859
and parent[0] == parent[1]
1860
and name[0] == name[1]
1861
and executable[0] == executable[1]):
1862
# Could happen when only the revision changed for a directory
1865
yield (file_id, (path_in_source, path_in_target), changed_content,
1866
versioned, parent, name, kind, executable)
1869
"""Return the number of entries in the inventory."""
1870
return len(self.id_to_entry)
1872
def _make_delta(self, old):
1873
"""Make an inventory delta from two inventories."""
1874
if type(old) != CHKInventory:
1875
return CommonInventory._make_delta(self, old)
1877
for key, old_value, self_value in \
1878
self.id_to_entry.iter_changes(old.id_to_entry):
1880
if old_value is not None:
1881
old_path = old.id2path(file_id)
1884
if self_value is not None:
1885
entry = self._bytes_to_entry(self_value)
1886
self._fileid_to_entry_cache[file_id] = entry
1887
new_path = self.id2path(file_id)
1891
delta.append((old_path, new_path, file_id, entry))
1894
def path2id(self, name):
1895
"""See CommonInventory.path2id()."""
1896
result = self._path_to_fileid_cache.get(name, None)
1898
result = CommonInventory.path2id(self, name)
1899
self._path_to_fileid_cache[name] = result
1903
"""Serialise the inventory to lines."""
1904
lines = ["chkinventory:\n"]
1905
if self._search_key_name != 'plain':
1906
# custom ordering grouping things that don't change together
1907
lines.append('search_key_name: %s\n' % (self._search_key_name,))
1908
lines.append("root_id: %s\n" % self.root_id)
1909
lines.append('parent_id_basename_to_file_id: %s\n' %
1910
self.parent_id_basename_to_file_id.key())
1911
lines.append("revision_id: %s\n" % self.revision_id)
1912
lines.append("id_to_entry: %s\n" % self.id_to_entry.key())
1914
lines.append("revision_id: %s\n" % self.revision_id)
1915
lines.append("root_id: %s\n" % self.root_id)
1916
if self.parent_id_basename_to_file_id is not None:
1917
lines.append('parent_id_basename_to_file_id: %s\n' %
1918
self.parent_id_basename_to_file_id.key())
1919
lines.append("id_to_entry: %s\n" % self.id_to_entry.key())
1924
"""Get the root entry."""
1925
return self[self.root_id]
1928
class CHKInventoryDirectory(InventoryDirectory):
1929
"""A directory in an inventory."""
1931
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
1932
'text_id', 'parent_id', '_children', 'executable',
1933
'revision', 'symlink_target', 'reference_revision',
1936
def __init__(self, file_id, name, parent_id, chk_inventory):
1937
# Don't call InventoryDirectory.__init__ - it isn't right for this
1939
InventoryEntry.__init__(self, file_id, name, parent_id)
1940
self._children = None
1941
self.kind = 'directory'
1942
self._chk_inventory = chk_inventory
1946
"""Access the list of children of this directory.
1948
With a parent_id_basename_to_file_id index, loads all the children,
1949
without loads the entire index. Without is bad. A more sophisticated
1950
proxy object might be nice, to allow partial loading of children as
1951
well when specific names are accessed. (So path traversal can be
1952
written in the obvious way but not examine siblings.).
1954
if self._children is not None:
1955
return self._children
1956
# No longer supported
1957
if self._chk_inventory.parent_id_basename_to_file_id is None:
1958
raise AssertionError("Inventories without"
1959
" parent_id_basename_to_file_id are no longer supported")
1961
# XXX: Todo - use proxy objects for the children rather than loading
1962
# all when the attribute is referenced.
1963
parent_id_index = self._chk_inventory.parent_id_basename_to_file_id
1965
for (parent_id, name_utf8), file_id in parent_id_index.iteritems(
1966
key_filter=[(self.file_id,)]):
1967
child_keys.add((file_id,))
1969
for file_id_key in child_keys:
1970
entry = self._chk_inventory._fileid_to_entry_cache.get(
1971
file_id_key[0], None)
1972
if entry is not None:
1973
result[entry.name] = entry
1974
cached.add(file_id_key)
1975
child_keys.difference_update(cached)
1976
# populate; todo: do by name
1977
id_to_entry = self._chk_inventory.id_to_entry
1978
for file_id_key, bytes in id_to_entry.iteritems(child_keys):
1979
entry = self._chk_inventory._bytes_to_entry(bytes)
1980
result[entry.name] = entry
1981
self._chk_inventory._fileid_to_entry_cache[file_id_key[0]] = entry
1982
self._children = result
1985
1395
entry_factory = {
1986
1396
'directory': InventoryDirectory,
1987
1397
'file': InventoryFile,
added
removed
bzrlib/inventory.py
