852
701
self.file_id, path)
854
703
def _forget_tree_state(self):
855
self.reference_revision = None
858
class Inventory(object):
859
"""Inventory of versioned files in a tree.
861
This describes which file_id is present at each point in the tree,
862
and possibly the SHA-1 or other information about the file.
863
Entries can be looked up either by path or by file_id.
865
The inventory represents a typical unix file tree, with
866
directories containing files and subdirectories. We never store
867
the full path to a file, because renaming a directory implicitly
868
moves all of its contents. This class internally maintains a
869
lookup tree that allows the children under a directory to be
872
InventoryEntry objects must not be modified after they are
873
inserted, other than through the Inventory API.
875
>>> inv = Inventory()
876
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
877
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
878
>>> inv['123-123'].name
881
May be treated as an iterator or set to look up file ids:
883
>>> bool(inv.path2id('hello.c'))
888
May also look up by name:
890
>>> [x[0] for x in inv.iter_entries()]
892
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
893
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
894
Traceback (most recent call last):
895
BzrError: parent_id {TREE_ROOT} not in inventory
896
>>> inv.add(InventoryFile('123-123', 'hello.c', 'TREE_ROOT-12345678-12345678'))
897
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT-12345678-12345678', sha1=None, len=None)
899
def __init__(self, root_id=ROOT_ID, revision_id=None):
900
"""Create or read an inventory.
902
If a working directory is specified, the inventory is read
903
from there. If the file is specified, read from that. If not,
904
the inventory is created empty.
906
The inventory is created with a default root directory, with
909
if root_id is not None:
910
assert root_id.__class__ == str
911
self._set_root(InventoryDirectory(root_id, u'', None))
915
self.revision_id = revision_id
917
def _set_root(self, ie):
919
self._byid = {self.root.file_id: self.root}
922
# TODO: jam 20051218 Should copy also copy the revision_id?
923
entries = self.iter_entries()
924
other = Inventory(entries.next()[1].file_id)
925
# copy recursively so we know directories will be added before
926
# their children. There are more efficient ways than this...
927
for path, entry in entries():
928
other.add(entry.copy())
932
return iter(self._byid)
935
"""Returns number of entries."""
936
return len(self._byid)
938
def iter_entries(self, from_dir=None):
939
"""Return (path, entry) pairs, in order by name."""
704
self.reference_revision = None
706
def _unchanged(self, previous_ie):
707
"""See InventoryEntry._unchanged."""
708
compatible = super(TreeReference, self)._unchanged(previous_ie)
709
if self.reference_revision != previous_ie.reference_revision:
714
class CommonInventory(object):
715
"""Basic inventory logic, defined in terms of primitives like has_id."""
717
def __contains__(self, file_id):
718
"""True if this entry contains a file with given id.
720
>>> inv = Inventory()
721
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
722
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
728
Note that this method along with __iter__ are not encouraged for use as
729
they are less clear than specific query methods - they may be rmeoved
732
return self.has_id(file_id)
734
def id2path(self, file_id):
735
"""Return as a string the path to file_id.
738
>>> e = i.add(InventoryDirectory('src-id', 'src', ROOT_ID))
739
>>> e = i.add(InventoryFile('foo-id', 'foo.c', parent_id='src-id'))
740
>>> print i.id2path('foo-id')
743
# get all names, skipping root
744
return '/'.join(reversed(
745
[parent.name for parent in
746
self._iter_file_id_parents(file_id)][:-1]))
748
def iter_entries(self, from_dir=None, recursive=True):
749
"""Return (path, entry) pairs, in order by name.
751
:param from_dir: if None, start from the root,
752
otherwise start from this directory (either file-id or entry)
753
:param recursive: recurse into directories or not
940
755
if from_dir is None:
941
756
if self.root is None:
943
758
from_dir = self.root
944
759
yield '', self.root
945
760
elif isinstance(from_dir, basestring):
946
from_dir = self._byid[from_dir]
761
from_dir = self[from_dir]
948
763
# unrolling the recursive called changed the time from
949
764
# 440ms/663ms (inline/total) to 116ms/116ms
950
765
children = from_dir.children.items()
768
for name, ie in children:
952
771
children = collections.deque(children)
953
772
stack = [(u'', children)]
1084
934
descend(child_ie, child_path)
1085
935
descend(self.root, u'')
938
def path2id(self, name):
939
"""Walk down through directories to return entry of last component.
941
names may be either a list of path components, or a single
942
string, in which case it is automatically split.
944
This returns the entry of the last component in the path,
945
which may be either a file or a directory.
947
Returns None IFF the path is not found.
949
if isinstance(name, basestring):
950
name = osutils.splitpath(name)
952
# mutter("lookup path %r" % name)
956
except errors.NoSuchId:
957
# root doesn't exist yet so nothing else can
963
children = getattr(parent, 'children', None)
972
return parent.file_id
974
def filter(self, specific_fileids):
975
"""Get an inventory view filtered against a set of file-ids.
977
Children of directories and parents are included.
979
The result may or may not reference the underlying inventory
980
so it should be treated as immutable.
982
interesting_parents = set()
983
for fileid in specific_fileids:
985
interesting_parents.update(self.get_idpath(fileid))
986
except errors.NoSuchId:
987
# This fileid is not in the inventory - that's ok
989
entries = self.iter_entries()
990
if self.root is None:
991
return Inventory(root_id=None)
992
other = Inventory(entries.next()[1].file_id)
993
other.root.revision = self.root.revision
994
other.revision_id = self.revision_id
995
directories_to_expand = set()
996
for path, entry in entries:
997
file_id = entry.file_id
998
if (file_id in specific_fileids
999
or entry.parent_id in directories_to_expand):
1000
if entry.kind == 'directory':
1001
directories_to_expand.add(file_id)
1002
elif file_id not in interesting_parents:
1004
other.add(entry.copy())
1007
def get_idpath(self, file_id):
1008
"""Return a list of file_ids for the path to an entry.
1010
The list contains one element for each directory followed by
1011
the id of the file itself. So the length of the returned list
1012
is equal to the depth of the file in the tree, counting the
1013
root directory as depth 1.
1016
for parent in self._iter_file_id_parents(file_id):
1017
p.insert(0, parent.file_id)
1021
class Inventory(CommonInventory):
1022
"""Inventory of versioned files in a tree.
1024
This describes which file_id is present at each point in the tree,
1025
and possibly the SHA-1 or other information about the file.
1026
Entries can be looked up either by path or by file_id.
1028
The inventory represents a typical unix file tree, with
1029
directories containing files and subdirectories. We never store
1030
the full path to a file, because renaming a directory implicitly
1031
moves all of its contents. This class internally maintains a
1032
lookup tree that allows the children under a directory to be
1035
InventoryEntry objects must not be modified after they are
1036
inserted, other than through the Inventory API.
1038
>>> inv = Inventory()
1039
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
1040
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
1041
>>> inv['123-123'].name
1044
May be treated as an iterator or set to look up file ids:
1046
>>> bool(inv.path2id('hello.c'))
1048
>>> '123-123' in inv
1051
May also look up by name:
1053
>>> [x[0] for x in inv.iter_entries()]
1055
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
1056
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
1057
Traceback (most recent call last):
1058
BzrError: parent_id {TREE_ROOT} not in inventory
1059
>>> inv.add(InventoryFile('123-123', 'hello.c', 'TREE_ROOT-12345678-12345678'))
1060
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT-12345678-12345678', sha1=None, len=None, revision=None)
1062
def __init__(self, root_id=ROOT_ID, revision_id=None):
1063
"""Create or read an inventory.
1065
If a working directory is specified, the inventory is read
1066
from there. If the file is specified, read from that. If not,
1067
the inventory is created empty.
1069
The inventory is created with a default root directory, with
1072
if root_id is not None:
1073
self._set_root(InventoryDirectory(root_id, u'', None))
1077
self.revision_id = revision_id
1080
# More than one page of ouput is not useful anymore to debug
1083
contents = repr(self._byid)
1084
if len(contents) > max_len:
1085
contents = contents[:(max_len-len(closing))] + closing
1086
return "<Inventory object at %x, contents=%r>" % (id(self), contents)
1088
def apply_delta(self, delta):
1089
"""Apply a delta to this inventory.
1091
See the inventory developers documentation for the theory behind
1094
:param delta: A list of changes to apply. After all the changes are
1095
applied the final inventory must be internally consistent, but it
1096
is ok to supply changes which, if only half-applied would have an
1097
invalid result - such as supplying two changes which rename two
1098
files, 'A' and 'B' with each other : [('A', 'B', 'A-id', a_entry),
1099
('B', 'A', 'B-id', b_entry)].
1101
Each change is a tuple, of the form (old_path, new_path, file_id,
1104
When new_path is None, the change indicates the removal of an entry
1105
from the inventory and new_entry will be ignored (using None is
1106
appropriate). If new_path is not None, then new_entry must be an
1107
InventoryEntry instance, which will be incorporated into the
1108
inventory (and replace any existing entry with the same file id).
1110
When old_path is None, the change indicates the addition of
1111
a new entry to the inventory.
1113
When neither new_path nor old_path are None, the change is a
1114
modification to an entry, such as a rename, reparent, kind change
1117
The children attribute of new_entry is ignored. This is because
1118
this method preserves children automatically across alterations to
1119
the parent of the children, and cases where the parent id of a
1120
child is changing require the child to be passed in as a separate
1121
change regardless. E.g. in the recursive deletion of a directory -
1122
the directory's children must be included in the delta, or the
1123
final inventory will be invalid.
1125
Note that a file_id must only appear once within a given delta.
1126
An AssertionError is raised otherwise.
1128
# Check that the delta is legal. It would be nice if this could be
1129
# done within the loops below but it's safer to validate the delta
1130
# before starting to mutate the inventory.
1131
unique_file_ids = set([f for _, _, f, _ in delta])
1132
if len(unique_file_ids) != len(delta):
1133
raise AssertionError("a file-id appears multiple times in %r"
1138
# Remove all affected items which were in the original inventory,
1139
# starting with the longest paths, thus ensuring parents are examined
1140
# after their children, which means that everything we examine has no
1141
# modified children remaining by the time we examine it.
1142
for old_path, file_id in sorted(((op, f) for op, np, f, e in delta
1143
if op is not None), reverse=True):
1144
if file_id not in self:
1147
# Preserve unaltered children of file_id for later reinsertion.
1148
file_id_children = getattr(self[file_id], 'children', {})
1149
if len(file_id_children):
1150
children[file_id] = file_id_children
1151
# Remove file_id and the unaltered children. If file_id is not
1152
# being deleted it will be reinserted back later.
1153
self.remove_recursive_id(file_id)
1154
# Insert all affected which should be in the new inventory, reattaching
1155
# their children if they had any. This is done from shortest path to
1156
# longest, ensuring that items which were modified and whose parents in
1157
# the resulting inventory were also modified, are inserted after their
1159
for new_path, new_entry in sorted((np, e) for op, np, f, e in
1160
delta if np is not None):
1161
if new_entry.kind == 'directory':
1162
# Pop the child which to allow detection of children whose
1163
# parents were deleted and which were not reattached to a new
1165
replacement = InventoryDirectory(new_entry.file_id,
1166
new_entry.name, new_entry.parent_id)
1167
replacement.revision = new_entry.revision
1168
replacement.children = children.pop(replacement.file_id, {})
1169
new_entry = replacement
1172
# Get the parent id that was deleted
1173
parent_id, children = children.popitem()
1174
raise errors.InconsistentDelta("<deleted>", parent_id,
1175
"The file id was deleted but its children were not deleted.")
1177
def _set_root(self, ie):
1179
self._byid = {self.root.file_id: self.root}
1182
# TODO: jam 20051218 Should copy also copy the revision_id?
1183
entries = self.iter_entries()
1184
if self.root is None:
1185
return Inventory(root_id=None)
1186
other = Inventory(entries.next()[1].file_id)
1187
other.root.revision = self.root.revision
1188
# copy recursively so we know directories will be added before
1189
# their children. There are more efficient ways than this...
1190
for path, entry in entries:
1191
other.add(entry.copy())
1194
def _get_mutable_inventory(self):
1195
"""See CommonInventory._get_mutable_inventory."""
1196
return copy.deepcopy(self)
1199
"""Iterate over all file-ids."""
1200
return iter(self._byid)
1202
def iter_just_entries(self):
1203
"""Iterate over all entries.
1088
def __contains__(self, file_id):
1089
"""True if this entry contains a file with given id.
1205
Unlike iter_entries(), just the entries are returned (not (path, ie))
1206
and the order of entries is undefined.
1091
>>> inv = Inventory()
1092
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1093
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
1208
XXX: We may not want to merge this into bzr.dev.
1099
file_id = osutils.safe_file_id(file_id)
1100
return (file_id in self._byid)
1210
if self.root is None:
1212
for _, ie in self._byid.iteritems():
1216
"""Returns number of entries."""
1217
return len(self._byid)
1102
1219
def __getitem__(self, file_id):
1103
1220
"""Return the entry for given file_id.
1105
1222
>>> inv = Inventory()
1106
1223
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
1107
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
1224
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
1108
1225
>>> inv['123123'].name
1111
file_id = osutils.safe_file_id(file_id)
1113
1229
return self._byid[file_id]
1114
1230
except KeyError:
1366
1437
del old_parent.children[file_ie.name]
1367
1438
new_parent.children[new_name] = file_ie
1369
1440
file_ie.name = new_name
1370
1441
file_ie.parent_id = new_parent_id
1372
1443
def is_root(self, file_id):
1373
file_id = osutils.safe_file_id(file_id)
1374
1444
return self.root is not None and file_id == self.root.file_id
1447
class CHKInventory(CommonInventory):
1448
"""An inventory persisted in a CHK store.
1450
By design, a CHKInventory is immutable so many of the methods
1451
supported by Inventory - add, rename, apply_delta, etc - are *not*
1452
supported. To create a new CHKInventory, use create_by_apply_delta()
1453
or from_inventory(), say.
1455
Internally, a CHKInventory has one or two CHKMaps:
1457
* id_to_entry - a map from (file_id,) => InventoryEntry as bytes
1458
* parent_id_basename_to_file_id - a map from (parent_id, basename_utf8)
1461
The second map is optional and not present in early CHkRepository's.
1463
No caching is performed: every method call or item access will perform
1464
requests to the storage layer. As such, keep references to objects you
1468
def __init__(self, search_key_name):
1469
CommonInventory.__init__(self)
1470
self._fileid_to_entry_cache = {}
1471
self._path_to_fileid_cache = {}
1472
self._search_key_name = search_key_name
1474
def __eq__(self, other):
1475
"""Compare two sets by comparing their contents."""
1476
if not isinstance(other, CHKInventory):
1477
return NotImplemented
1479
this_key = self.id_to_entry.key()
1480
other_key = other.id_to_entry.key()
1481
this_pid_key = self.parent_id_basename_to_file_id.key()
1482
other_pid_key = other.parent_id_basename_to_file_id.key()
1483
if None in (this_key, this_pid_key, other_key, other_pid_key):
1485
return this_key == other_key and this_pid_key == other_pid_key
1487
def _entry_to_bytes(self, entry):
1488
"""Serialise entry as a single bytestring.
1490
:param Entry: An inventory entry.
1491
:return: A bytestring for the entry.
1494
ENTRY ::= FILE | DIR | SYMLINK | TREE
1495
FILE ::= "file: " COMMON SEP SHA SEP SIZE SEP EXECUTABLE
1496
DIR ::= "dir: " COMMON
1497
SYMLINK ::= "symlink: " COMMON SEP TARGET_UTF8
1498
TREE ::= "tree: " COMMON REFERENCE_REVISION
1499
COMMON ::= FILE_ID SEP PARENT_ID SEP NAME_UTF8 SEP REVISION
1502
if entry.parent_id is not None:
1503
parent_str = entry.parent_id
1506
name_str = entry.name.encode("utf8")
1507
if entry.kind == 'file':
1508
if entry.executable:
1512
return "file: %s\n%s\n%s\n%s\n%s\n%d\n%s" % (
1513
entry.file_id, parent_str, name_str, entry.revision,
1514
entry.text_sha1, entry.text_size, exec_str)
1515
elif entry.kind == 'directory':
1516
return "dir: %s\n%s\n%s\n%s" % (
1517
entry.file_id, parent_str, name_str, entry.revision)
1518
elif entry.kind == 'symlink':
1519
return "symlink: %s\n%s\n%s\n%s\n%s" % (
1520
entry.file_id, parent_str, name_str, entry.revision,
1521
entry.symlink_target.encode("utf8"))
1522
elif entry.kind == 'tree-reference':
1523
return "tree: %s\n%s\n%s\n%s\n%s" % (
1524
entry.file_id, parent_str, name_str, entry.revision,
1525
entry.reference_revision)
1527
raise ValueError("unknown kind %r" % entry.kind)
1530
def _bytes_to_utf8name_key(bytes):
1531
"""Get the file_id, revision_id key out of bytes."""
1532
# We don't normally care about name, except for times when we want
1533
# to filter out empty names because of non rich-root...
1534
sections = bytes.split('\n')
1535
kind, file_id = sections[0].split(': ')
1536
return (sections[2], file_id, sections[3])
1538
def _bytes_to_entry(self, bytes):
1539
"""Deserialise a serialised entry."""
1540
sections = bytes.split('\n')
1541
if sections[0].startswith("file: "):
1542
result = InventoryFile(sections[0][6:],
1543
sections[2].decode('utf8'),
1545
result.text_sha1 = sections[4]
1546
result.text_size = int(sections[5])
1547
result.executable = sections[6] == "Y"
1548
elif sections[0].startswith("dir: "):
1549
result = CHKInventoryDirectory(sections[0][5:],
1550
sections[2].decode('utf8'),
1552
elif sections[0].startswith("symlink: "):
1553
result = InventoryLink(sections[0][9:],
1554
sections[2].decode('utf8'),
1556
result.symlink_target = sections[4].decode('utf8')
1557
elif sections[0].startswith("tree: "):
1558
result = TreeReference(sections[0][6:],
1559
sections[2].decode('utf8'),
1561
result.reference_revision = sections[4]
1563
raise ValueError("Not a serialised entry %r" % bytes)
1564
result.revision = sections[3]
1565
if result.parent_id == '':
1566
result.parent_id = None
1567
self._fileid_to_entry_cache[result.file_id] = result
1570
def _get_mutable_inventory(self):
1571
"""See CommonInventory._get_mutable_inventory."""
1572
entries = self.iter_entries()
1573
inv = Inventory(None, self.revision_id)
1574
for path, inv_entry in entries:
1575
inv.add(inv_entry.copy())
1578
def create_by_apply_delta(self, inventory_delta, new_revision_id,
1579
propagate_caches=False):
1580
"""Create a new CHKInventory by applying inventory_delta to this one.
1582
See the inventory developers documentation for the theory behind
1585
:param inventory_delta: The inventory delta to apply. See
1586
Inventory.apply_delta for details.
1587
:param new_revision_id: The revision id of the resulting CHKInventory.
1588
:param propagate_caches: If True, the caches for this inventory are
1589
copied to and updated for the result.
1590
:return: The new CHKInventory.
1592
result = CHKInventory(self._search_key_name)
1593
if propagate_caches:
1594
# Just propagate the path-to-fileid cache for now
1595
result._path_to_fileid_cache = dict(self._path_to_fileid_cache.iteritems())
1596
search_key_func = chk_map.search_key_registry.get(self._search_key_name)
1597
self.id_to_entry._ensure_root()
1598
maximum_size = self.id_to_entry._root_node.maximum_size
1599
result.revision_id = new_revision_id
1600
result.id_to_entry = chk_map.CHKMap(
1601
self.id_to_entry._store,
1602
self.id_to_entry.key(),
1603
search_key_func=search_key_func)
1604
result.id_to_entry._ensure_root()
1605
result.id_to_entry._root_node.set_maximum_size(maximum_size)
1606
parent_id_basename_delta = []
1607
if self.parent_id_basename_to_file_id is not None:
1608
result.parent_id_basename_to_file_id = chk_map.CHKMap(
1609
self.parent_id_basename_to_file_id._store,
1610
self.parent_id_basename_to_file_id.key(),
1611
search_key_func=search_key_func)
1612
result.parent_id_basename_to_file_id._ensure_root()
1613
self.parent_id_basename_to_file_id._ensure_root()
1614
result_p_id_root = result.parent_id_basename_to_file_id._root_node
1615
p_id_root = self.parent_id_basename_to_file_id._root_node
1616
result_p_id_root.set_maximum_size(p_id_root.maximum_size)
1617
result_p_id_root._key_width = p_id_root._key_width
1619
result.parent_id_basename_to_file_id = None
1620
result.root_id = self.root_id
1621
id_to_entry_delta = []
1622
for old_path, new_path, file_id, entry in inventory_delta:
1625
result.root_id = file_id
1626
if new_path is None:
1631
if propagate_caches:
1633
del result._path_to_fileid_cache[old_path]
1637
new_key = (file_id,)
1638
new_value = result._entry_to_bytes(entry)
1639
# Update caches. It's worth doing this whether
1640
# we're propagating the old caches or not.
1641
result._path_to_fileid_cache[new_path] = file_id
1642
if old_path is None:
1645
old_key = (file_id,)
1646
id_to_entry_delta.append((old_key, new_key, new_value))
1647
if result.parent_id_basename_to_file_id is not None:
1648
# parent_id, basename changes
1649
if old_path is None:
1652
old_entry = self[file_id]
1653
old_key = self._parent_id_basename_key(old_entry)
1654
if new_path is None:
1658
new_key = self._parent_id_basename_key(entry)
1660
if old_key != new_key:
1661
# If the two keys are the same, the value will be unchanged
1662
# as its always the file id.
1663
parent_id_basename_delta.append((old_key, new_key, new_value))
1664
result.id_to_entry.apply_delta(id_to_entry_delta)
1665
if parent_id_basename_delta:
1666
result.parent_id_basename_to_file_id.apply_delta(parent_id_basename_delta)
1670
def deserialise(klass, chk_store, bytes, expected_revision_id):
1671
"""Deserialise a CHKInventory.
1673
:param chk_store: A CHK capable VersionedFiles instance.
1674
:param bytes: The serialised bytes.
1675
:param expected_revision_id: The revision ID we think this inventory is
1677
:return: A CHKInventory
1679
lines = bytes.split('\n')
1681
raise AssertionError('bytes to deserialize must end with an eol')
1683
if lines[0] != 'chkinventory:':
1684
raise ValueError("not a serialised CHKInventory: %r" % bytes)
1686
allowed_keys = frozenset(['root_id', 'revision_id', 'search_key_name',
1687
'parent_id_basename_to_file_id',
1689
for line in lines[1:]:
1690
key, value = line.split(': ', 1)
1691
if key not in allowed_keys:
1692
raise errors.BzrError('Unknown key in inventory: %r\n%r'
1695
raise errors.BzrError('Duplicate key in inventory: %r\n%r'
1698
revision_id = info['revision_id']
1699
root_id = info['root_id']
1700
search_key_name = info.get('search_key_name', 'plain')
1701
parent_id_basename_to_file_id = info.get(
1702
'parent_id_basename_to_file_id', None)
1703
id_to_entry = info['id_to_entry']
1705
result = CHKInventory(search_key_name)
1706
result.revision_id = revision_id
1707
result.root_id = root_id
1708
search_key_func = chk_map.search_key_registry.get(
1709
result._search_key_name)
1710
if parent_id_basename_to_file_id is not None:
1711
result.parent_id_basename_to_file_id = chk_map.CHKMap(
1712
chk_store, (parent_id_basename_to_file_id,),
1713
search_key_func=search_key_func)
1715
result.parent_id_basename_to_file_id = None
1717
result.id_to_entry = chk_map.CHKMap(chk_store, (id_to_entry,),
1718
search_key_func=search_key_func)
1719
if (result.revision_id,) != expected_revision_id:
1720
raise ValueError("Mismatched revision id and expected: %r, %r" %
1721
(result.revision_id, expected_revision_id))
1725
def from_inventory(klass, chk_store, inventory, maximum_size=0, search_key_name='plain'):
1726
"""Create a CHKInventory from an existing inventory.
1728
The content of inventory is copied into the chk_store, and a
1729
CHKInventory referencing that is returned.
1731
:param chk_store: A CHK capable VersionedFiles instance.
1732
:param inventory: The inventory to copy.
1733
:param maximum_size: The CHKMap node size limit.
1734
:param search_key_name: The identifier for the search key function
1736
result = klass(search_key_name)
1737
result.revision_id = inventory.revision_id
1738
result.root_id = inventory.root.file_id
1740
entry_to_bytes = result._entry_to_bytes
1741
parent_id_basename_key = result._parent_id_basename_key
1742
id_to_entry_dict = {}
1743
parent_id_basename_dict = {}
1744
for path, entry in inventory.iter_entries():
1745
id_to_entry_dict[(entry.file_id,)] = entry_to_bytes(entry)
1746
p_id_key = parent_id_basename_key(entry)
1747
parent_id_basename_dict[p_id_key] = entry.file_id
1749
result._populate_from_dicts(chk_store, id_to_entry_dict,
1750
parent_id_basename_dict, maximum_size=maximum_size)
1753
def _populate_from_dicts(self, chk_store, id_to_entry_dict,
1754
parent_id_basename_dict, maximum_size):
1755
search_key_func = chk_map.search_key_registry.get(self._search_key_name)
1756
root_key = chk_map.CHKMap.from_dict(chk_store, id_to_entry_dict,
1757
maximum_size=maximum_size, key_width=1,
1758
search_key_func=search_key_func)
1759
self.id_to_entry = chk_map.CHKMap(chk_store, root_key,
1761
root_key = chk_map.CHKMap.from_dict(chk_store,
1762
parent_id_basename_dict,
1763
maximum_size=maximum_size, key_width=2,
1764
search_key_func=search_key_func)
1765
self.parent_id_basename_to_file_id = chk_map.CHKMap(chk_store,
1766
root_key, search_key_func)
1768
def _parent_id_basename_key(self, entry):
1769
"""Create a key for a entry in a parent_id_basename_to_file_id index."""
1770
if entry.parent_id is not None:
1771
parent_id = entry.parent_id
1774
return parent_id, entry.name.encode('utf8')
1776
def __getitem__(self, file_id):
1777
"""map a single file_id -> InventoryEntry."""
1779
raise errors.NoSuchId(self, file_id)
1780
result = self._fileid_to_entry_cache.get(file_id, None)
1781
if result is not None:
1784
return self._bytes_to_entry(
1785
self.id_to_entry.iteritems([(file_id,)]).next()[1])
1786
except StopIteration:
1787
# really we're passing an inventory, not a tree...
1788
raise errors.NoSuchId(self, file_id)
1790
def has_id(self, file_id):
1791
# Perhaps have an explicit 'contains' method on CHKMap ?
1792
if self._fileid_to_entry_cache.get(file_id, None) is not None:
1794
return len(list(self.id_to_entry.iteritems([(file_id,)]))) == 1
1796
def is_root(self, file_id):
1797
return file_id == self.root_id
1799
def _iter_file_id_parents(self, file_id):
1800
"""Yield the parents of file_id up to the root."""
1801
while file_id is not None:
1805
raise errors.NoSuchId(tree=self, file_id=file_id)
1807
file_id = ie.parent_id
1810
"""Iterate over all file-ids."""
1811
for key, _ in self.id_to_entry.iteritems():
1814
def iter_just_entries(self):
1815
"""Iterate over all entries.
1817
Unlike iter_entries(), just the entries are returned (not (path, ie))
1818
and the order of entries is undefined.
1820
XXX: We may not want to merge this into bzr.dev.
1822
for key, entry in self.id_to_entry.iteritems():
1824
ie = self._fileid_to_entry_cache.get(file_id, None)
1826
ie = self._bytes_to_entry(entry)
1827
self._fileid_to_entry_cache[file_id] = ie
1830
def iter_changes(self, basis):
1831
"""Generate a Tree.iter_changes change list between this and basis.
1833
:param basis: Another CHKInventory.
1834
:return: An iterator over the changes between self and basis, as per
1835
tree.iter_changes().
1837
# We want: (file_id, (path_in_source, path_in_target),
1838
# changed_content, versioned, parent, name, kind,
1840
for key, basis_value, self_value in \
1841
self.id_to_entry.iter_changes(basis.id_to_entry):
1843
if basis_value is not None:
1844
basis_entry = basis._bytes_to_entry(basis_value)
1845
path_in_source = basis.id2path(file_id)
1846
basis_parent = basis_entry.parent_id
1847
basis_name = basis_entry.name
1848
basis_executable = basis_entry.executable
1850
path_in_source = None
1853
basis_executable = None
1854
if self_value is not None:
1855
self_entry = self._bytes_to_entry(self_value)
1856
path_in_target = self.id2path(file_id)
1857
self_parent = self_entry.parent_id
1858
self_name = self_entry.name
1859
self_executable = self_entry.executable
1861
path_in_target = None
1864
self_executable = None
1865
if basis_value is None:
1867
kind = (None, self_entry.kind)
1868
versioned = (False, True)
1869
elif self_value is None:
1871
kind = (basis_entry.kind, None)
1872
versioned = (True, False)
1874
kind = (basis_entry.kind, self_entry.kind)
1875
versioned = (True, True)
1876
changed_content = False
1877
if kind[0] != kind[1]:
1878
changed_content = True
1879
elif kind[0] == 'file':
1880
if (self_entry.text_size != basis_entry.text_size or
1881
self_entry.text_sha1 != basis_entry.text_sha1):
1882
changed_content = True
1883
elif kind[0] == 'symlink':
1884
if self_entry.symlink_target != basis_entry.symlink_target:
1885
changed_content = True
1886
elif kind[0] == 'tree-reference':
1887
if (self_entry.reference_revision !=
1888
basis_entry.reference_revision):
1889
changed_content = True
1890
parent = (basis_parent, self_parent)
1891
name = (basis_name, self_name)
1892
executable = (basis_executable, self_executable)
1893
if (not changed_content
1894
and parent[0] == parent[1]
1895
and name[0] == name[1]
1896
and executable[0] == executable[1]):
1897
# Could happen when only the revision changed for a directory
1900
yield (file_id, (path_in_source, path_in_target), changed_content,
1901
versioned, parent, name, kind, executable)
1904
"""Return the number of entries in the inventory."""
1905
return len(self.id_to_entry)
1907
def _make_delta(self, old):
1908
"""Make an inventory delta from two inventories."""
1909
if type(old) != CHKInventory:
1910
return CommonInventory._make_delta(self, old)
1912
for key, old_value, self_value in \
1913
self.id_to_entry.iter_changes(old.id_to_entry):
1915
if old_value is not None:
1916
old_path = old.id2path(file_id)
1919
if self_value is not None:
1920
entry = self._bytes_to_entry(self_value)
1921
self._fileid_to_entry_cache[file_id] = entry
1922
new_path = self.id2path(file_id)
1926
delta.append((old_path, new_path, file_id, entry))
1929
def path2id(self, name):
1930
"""See CommonInventory.path2id()."""
1931
result = self._path_to_fileid_cache.get(name, None)
1933
result = CommonInventory.path2id(self, name)
1934
self._path_to_fileid_cache[name] = result
1938
"""Serialise the inventory to lines."""
1939
lines = ["chkinventory:\n"]
1940
if self._search_key_name != 'plain':
1941
# custom ordering grouping things that don't change together
1942
lines.append('search_key_name: %s\n' % (self._search_key_name,))
1943
lines.append("root_id: %s\n" % self.root_id)
1944
lines.append('parent_id_basename_to_file_id: %s\n' %
1945
self.parent_id_basename_to_file_id.key())
1946
lines.append("revision_id: %s\n" % self.revision_id)
1947
lines.append("id_to_entry: %s\n" % self.id_to_entry.key())
1949
lines.append("revision_id: %s\n" % self.revision_id)
1950
lines.append("root_id: %s\n" % self.root_id)
1951
if self.parent_id_basename_to_file_id is not None:
1952
lines.append('parent_id_basename_to_file_id: %s\n' %
1953
self.parent_id_basename_to_file_id.key())
1954
lines.append("id_to_entry: %s\n" % self.id_to_entry.key())
1959
"""Get the root entry."""
1960
return self[self.root_id]
1963
class CHKInventoryDirectory(InventoryDirectory):
1964
"""A directory in an inventory."""
1966
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
1967
'text_id', 'parent_id', '_children', 'executable',
1968
'revision', 'symlink_target', 'reference_revision',
1971
def __init__(self, file_id, name, parent_id, chk_inventory):
1972
# Don't call InventoryDirectory.__init__ - it isn't right for this
1974
InventoryEntry.__init__(self, file_id, name, parent_id)
1975
self._children = None
1976
self.kind = 'directory'
1977
self._chk_inventory = chk_inventory
1981
"""Access the list of children of this directory.
1983
With a parent_id_basename_to_file_id index, loads all the children,
1984
without loads the entire index. Without is bad. A more sophisticated
1985
proxy object might be nice, to allow partial loading of children as
1986
well when specific names are accessed. (So path traversal can be
1987
written in the obvious way but not examine siblings.).
1989
if self._children is not None:
1990
return self._children
1991
# No longer supported
1992
if self._chk_inventory.parent_id_basename_to_file_id is None:
1993
raise AssertionError("Inventories without"
1994
" parent_id_basename_to_file_id are no longer supported")
1996
# XXX: Todo - use proxy objects for the children rather than loading
1997
# all when the attribute is referenced.
1998
parent_id_index = self._chk_inventory.parent_id_basename_to_file_id
2000
for (parent_id, name_utf8), file_id in parent_id_index.iteritems(
2001
key_filter=[(self.file_id,)]):
2002
child_keys.add((file_id,))
2004
for file_id_key in child_keys:
2005
entry = self._chk_inventory._fileid_to_entry_cache.get(
2006
file_id_key[0], None)
2007
if entry is not None:
2008
result[entry.name] = entry
2009
cached.add(file_id_key)
2010
child_keys.difference_update(cached)
2011
# populate; todo: do by name
2012
id_to_entry = self._chk_inventory.id_to_entry
2013
for file_id_key, bytes in id_to_entry.iteritems(child_keys):
2014
entry = self._chk_inventory._bytes_to_entry(bytes)
2015
result[entry.name] = entry
2016
self._chk_inventory._fileid_to_entry_cache[file_id_key[0]] = entry
2017
self._children = result
1377
2020
entry_factory = {
1378
2021
'directory': InventoryDirectory,
1379
2022
'file': InventoryFile,
added
removed
bzrlib/inventory.py
