273
272
:param rev_id: Revision id from which this InventoryEntry was loaded.
274
273
Not necessarily the last-changed revision for this file.
275
274
:param inv: Inventory from which the entry was loaded.
275
:param tree: RevisionTree for this entry.
277
277
if self.parent_id is not None:
278
278
if not inv.has_id(self.parent_id):
279
279
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
280
280
% (self.parent_id, rev_id))
281
checker._add_entry_to_text_key_references(inv, self)
282
self._check(checker, rev_id)
281
self._check(checker, rev_id, tree)
284
def _check(self, checker, rev_id):
283
def _check(self, checker, rev_id, tree):
285
284
"""Check this inventory entry for kind specific errors."""
286
checker._report_items.append(
287
'unknown entry kind %r in revision {%s}' % (self.kind, rev_id))
285
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
290
289
"""Clone this inventory entry."""
432
431
'text_id', 'parent_id', 'children', 'executable',
433
432
'revision', 'symlink_target', 'reference_revision']
435
def _check(self, checker, rev_id):
434
def _check(self, checker, rev_id, tree):
436
435
"""See InventoryEntry._check"""
437
if (self.text_sha1 is not None or self.text_size is not None or
438
self.text_id is not None):
439
checker._report_items.append('directory {%s} has text in revision {%s}'
436
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
437
raise BzrCheckError('directory {%s} has text in revision {%s}'
440
438
% (self.file_id, rev_id))
441
# In non rich root repositories we do not expect a file graph for the
443
if self.name == '' and not checker.rich_roots:
445
# Directories are stored as an empty file, but the file should exist
446
# to provide a per-fileid log. The hash of every directory content is
447
# "da..." below (the sha1sum of '').
448
checker.add_pending_item(rev_id,
449
('texts', self.file_id, self.revision), 'text',
450
'da39a3ee5e6b4b0d3255bfef95601890afd80709')
453
441
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
486
474
'text_id', 'parent_id', 'children', 'executable',
487
475
'revision', 'symlink_target', 'reference_revision']
489
def _check(self, checker, tree_revision_id):
477
def _check(self, checker, tree_revision_id, tree):
490
478
"""See InventoryEntry._check"""
491
# TODO: check size too.
492
checker.add_pending_item(tree_revision_id,
493
('texts', self.file_id, self.revision), 'text',
495
if self.text_size is None:
496
checker._report_items.append(
497
'fileid {%s} in {%s} has None for text_size' % (self.file_id,
479
key = (self.file_id, self.revision)
480
if key in checker.checked_texts:
481
prev_sha = checker.checked_texts[key]
482
if prev_sha != self.text_sha1:
484
'mismatched sha1 on {%s} in {%s} (%s != %s) %r' %
485
(self.file_id, tree_revision_id, prev_sha, self.text_sha1,
488
checker.repeated_text_cnt += 1
491
checker.checked_text_cnt += 1
492
# We can't check the length, because Weave doesn't store that
493
# information, and the whole point of looking at the weave's
494
# sha1sum is that we don't have to extract the text.
495
if (self.text_sha1 != tree._repository.texts.get_sha1s([key])[key]):
496
raise BzrCheckError('text {%s} version {%s} wrong sha1' % key)
497
checker.checked_texts[key] = self.text_sha1
501
500
other = InventoryFile(self.file_id, self.name, self.parent_id)
599
598
'text_id', 'parent_id', 'children', 'executable',
600
599
'revision', 'symlink_target', 'reference_revision']
602
def _check(self, checker, tree_revision_id):
601
def _check(self, checker, rev_id, tree):
603
602
"""See InventoryEntry._check"""
604
603
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
605
checker._report_items.append(
606
'symlink {%s} has text in revision {%s}'
607
% (self.file_id, tree_revision_id))
604
raise BzrCheckError('symlink {%s} has text in revision {%s}'
605
% (self.file_id, rev_id))
608
606
if self.symlink_target is None:
609
checker._report_items.append(
610
'symlink {%s} has no target in revision {%s}'
611
% (self.file_id, tree_revision_id))
612
# Symlinks are stored as ''
613
checker.add_pending_item(tree_revision_id,
614
('texts', self.file_id, self.revision), 'text',
615
'da39a3ee5e6b4b0d3255bfef95601890afd80709')
607
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
608
% (self.file_id, rev_id))
618
611
other = InventoryLink(self.file_id, self.name, self.parent_id)
959
947
descend(self.root, u'')
962
def path2id(self, relpath):
950
def path2id(self, name):
963
951
"""Walk down through directories to return entry of last component.
965
:param relpath: may be either a list of path components, or a single
966
string, in which case it is automatically split.
953
names may be either a list of path components, or a single
954
string, in which case it is automatically split.
968
956
This returns the entry of the last component in the path,
969
957
which may be either a file or a directory.
971
959
Returns None IFF the path is not found.
973
if isinstance(relpath, basestring):
974
names = osutils.splitpath(relpath)
961
if isinstance(name, basestring):
962
name = osutils.splitpath(name)
964
# mutter("lookup path %r" % name)
979
967
parent = self.root
1155
1138
# modified children remaining by the time we examine it.
1156
1139
for old_path, file_id in sorted(((op, f) for op, np, f, e in delta
1157
1140
if op is not None), reverse=True):
1141
if file_id not in self:
1158
1144
# Preserve unaltered children of file_id for later reinsertion.
1159
1145
file_id_children = getattr(self[file_id], 'children', {})
1160
1146
if len(file_id_children):
1161
1147
children[file_id] = file_id_children
1162
if self.id2path(file_id) != old_path:
1163
raise errors.InconsistentDelta(old_path, file_id,
1164
"Entry was at wrong other path %r." % self.id2path(file_id))
1165
1148
# Remove file_id and the unaltered children. If file_id is not
1166
1149
# being deleted it will be reinserted back later.
1167
1150
self.remove_recursive_id(file_id)
1183
1166
new_entry = replacement
1185
1168
self.add(new_entry)
1186
except errors.DuplicateFileId:
1187
raise errors.InconsistentDelta(new_path, new_entry.file_id,
1188
"New id is already present in target.")
1189
1169
except AttributeError:
1190
1170
raise errors.InconsistentDelta(new_path, new_entry.file_id,
1191
1171
"Parent is not a directory.")
1192
if self.id2path(new_entry.file_id) != new_path:
1193
raise errors.InconsistentDelta(new_path, new_entry.file_id,
1194
"New path is not consistent with parent path.")
1195
1172
if len(children):
1196
1173
# Get the parent id that was deleted
1197
1174
parent_id, children = children.popitem()
1198
1175
raise errors.InconsistentDelta("<deleted>", parent_id,
1199
1176
"The file id was deleted but its children were not deleted.")
1201
def create_by_apply_delta(self, inventory_delta, new_revision_id,
1202
propagate_caches=False):
1203
"""See CHKInventory.create_by_apply_delta()"""
1204
new_inv = self.copy()
1205
new_inv.apply_delta(inventory_delta)
1206
new_inv.revision_id = new_revision_id
1209
1178
def _set_root(self, ie):
1211
1180
self._byid = {self.root.file_id: self.root}
1556
1528
raise ValueError("unknown kind %r" % entry.kind)
1558
def _expand_fileids_to_parents_and_children(self, file_ids):
1559
"""Give a more wholistic view starting with the given file_ids.
1561
For any file_id which maps to a directory, we will include all children
1562
of that directory. We will also include all directories which are
1563
parents of the given file_ids, but we will not include their children.
1570
fringle # fringle-id
1574
if given [foo-id] we will include
1575
TREE_ROOT as interesting parents
1577
foo-id, baz-id, frob-id, fringle-id
1581
# TODO: Pre-pass over the list of fileids to see if anything is already
1582
# deserialized in self._fileid_to_entry_cache
1584
directories_to_expand = set()
1585
children_of_parent_id = {}
1586
# It is okay if some of the fileids are missing
1587
for entry in self._getitems(file_ids):
1588
if entry.kind == 'directory':
1589
directories_to_expand.add(entry.file_id)
1590
interesting.add(entry.parent_id)
1591
children_of_parent_id.setdefault(entry.parent_id, []
1592
).append(entry.file_id)
1594
# Now, interesting has all of the direct parents, but not the
1595
# parents of those parents. It also may have some duplicates with
1597
remaining_parents = interesting.difference(file_ids)
1598
# When we hit the TREE_ROOT, we'll get an interesting parent of None,
1599
# but we don't actually want to recurse into that
1600
interesting.add(None) # this will auto-filter it in the loop
1601
remaining_parents.discard(None)
1602
while remaining_parents:
1603
next_parents = set()
1604
for entry in self._getitems(remaining_parents):
1605
next_parents.add(entry.parent_id)
1606
children_of_parent_id.setdefault(entry.parent_id, []
1607
).append(entry.file_id)
1608
# Remove any search tips we've already processed
1609
remaining_parents = next_parents.difference(interesting)
1610
interesting.update(remaining_parents)
1611
# We should probably also .difference(directories_to_expand)
1612
interesting.update(file_ids)
1613
interesting.discard(None)
1614
while directories_to_expand:
1615
# Expand directories by looking in the
1616
# parent_id_basename_to_file_id map
1617
keys = [StaticTuple(f,).intern() for f in directories_to_expand]
1618
directories_to_expand = set()
1619
items = self.parent_id_basename_to_file_id.iteritems(keys)
1620
next_file_ids = set([item[1] for item in items])
1621
next_file_ids = next_file_ids.difference(interesting)
1622
interesting.update(next_file_ids)
1623
for entry in self._getitems(next_file_ids):
1624
if entry.kind == 'directory':
1625
directories_to_expand.add(entry.file_id)
1626
children_of_parent_id.setdefault(entry.parent_id, []
1627
).append(entry.file_id)
1628
return interesting, children_of_parent_id
1630
def filter(self, specific_fileids):
1631
"""Get an inventory view filtered against a set of file-ids.
1633
Children of directories and parents are included.
1635
The result may or may not reference the underlying inventory
1636
so it should be treated as immutable.
1639
parent_to_children) = self._expand_fileids_to_parents_and_children(
1641
# There is some overlap here, but we assume that all interesting items
1642
# are in the _fileid_to_entry_cache because we had to read them to
1643
# determine if they were a dir we wanted to recurse, or just a file
1644
# This should give us all the entries we'll want to add, so start
1646
other = Inventory(self.root_id)
1647
other.root.revision = self.root.revision
1648
other.revision_id = self.revision_id
1649
if not interesting or not parent_to_children:
1650
# empty filter, or filtering entrys that don't exist
1651
# (if even 1 existed, then we would have populated
1652
# parent_to_children with at least the tree root.)
1654
cache = self._fileid_to_entry_cache
1656
remaining_children = collections.deque(parent_to_children[self.root_id])
1658
import pdb; pdb.set_trace()
1660
while remaining_children:
1661
file_id = remaining_children.popleft()
1663
if ie.kind == 'directory':
1664
ie = ie.copy() # We create a copy to depopulate the .children attribute
1665
# TODO: depending on the uses of 'other' we should probably alwyas
1666
# '.copy()' to prevent someone from mutating other and
1667
# invaliding our internal cache
1669
if file_id in parent_to_children:
1670
remaining_children.extend(parent_to_children[file_id])
1674
1531
def _bytes_to_utf8name_key(bytes):
1675
1532
"""Get the file_id, revision_id key out of bytes."""
1749
1604
search_key_func=search_key_func)
1750
1605
result.id_to_entry._ensure_root()
1751
1606
result.id_to_entry._root_node.set_maximum_size(maximum_size)
1752
# Change to apply to the parent_id_basename delta. The dict maps
1753
# (parent_id, basename) -> (old_key, new_value). We use a dict because
1754
# when a path has its id replaced (e.g. the root is changed, or someone
1755
# does bzr mv a b, bzr mv c a, we should output a single change to this
1756
# map rather than two.
1757
parent_id_basename_delta = {}
1607
parent_id_basename_delta = []
1758
1608
if self.parent_id_basename_to_file_id is not None:
1759
1609
result.parent_id_basename_to_file_id = chk_map.CHKMap(
1760
1610
self.parent_id_basename_to_file_id._store,
1780
1630
inventory_delta = _check_delta_unique_new_paths(inventory_delta)
1781
1631
# Check for entries that don't match the fileid
1782
1632
inventory_delta = _check_delta_ids_match_entry(inventory_delta)
1783
# Check for nonsense fileids
1784
inventory_delta = _check_delta_ids_are_valid(inventory_delta)
1785
# Check for new_path <-> entry consistency
1786
inventory_delta = _check_delta_new_path_entry_both_or_None(
1788
# All changed entries need to have their parents be directories and be
1789
# at the right path. This set contains (path, id) tuples.
1633
# All changed entries need to have their parents be directories.
1790
1634
parents = set()
1791
# When we delete an item, all the children of it must be either deleted
1792
# or altered in their own right. As we batch process the change via
1793
# CHKMap.apply_delta, we build a set of things to use to validate the
1797
1635
for old_path, new_path, file_id, entry in inventory_delta:
1798
1636
# file id changes
1799
1637
if new_path == '':
1808
1646
del result._path_to_fileid_cache[old_path]
1809
1647
except KeyError:
1811
deletes.add(file_id)
1813
new_key = StaticTuple(file_id,)
1650
new_key = (file_id,)
1814
1651
new_value = result._entry_to_bytes(entry)
1815
1652
# Update caches. It's worth doing this whether
1816
1653
# we're propagating the old caches or not.
1817
1654
result._path_to_fileid_cache[new_path] = file_id
1818
parents.add((split(new_path)[0], entry.parent_id))
1655
parents.add(entry.parent_id)
1819
1656
if old_path is None:
1822
old_key = StaticTuple(file_id,)
1823
if self.id2path(file_id) != old_path:
1824
raise errors.InconsistentDelta(old_path, file_id,
1825
"Entry was at wrong other path %r." %
1826
self.id2path(file_id))
1827
altered.add(file_id)
1828
id_to_entry_delta.append(StaticTuple(old_key, new_key, new_value))
1659
old_key = (file_id,)
1660
id_to_entry_delta.append((old_key, new_key, new_value))
1829
1661
if result.parent_id_basename_to_file_id is not None:
1830
1662
# parent_id, basename changes
1831
1663
if old_path is None:
1840
1672
new_key = self._parent_id_basename_key(entry)
1841
1673
new_value = file_id
1842
# If the two keys are the same, the value will be unchanged
1843
# as its always the file id for this entry.
1844
1674
if old_key != new_key:
1845
# Transform a change into explicit delete/add preserving
1846
# a possible match on the key from a different file id.
1847
if old_key is not None:
1848
parent_id_basename_delta.setdefault(
1849
old_key, [None, None])[0] = old_key
1850
if new_key is not None:
1851
parent_id_basename_delta.setdefault(
1852
new_key, [None, None])[1] = new_value
1853
# validate that deletes are complete.
1854
for file_id in deletes:
1855
entry = self[file_id]
1856
if entry.kind != 'directory':
1858
# This loop could potentially be better by using the id_basename
1859
# map to just get the child file ids.
1860
for child in entry.children.values():
1861
if child.file_id not in altered:
1862
raise errors.InconsistentDelta(self.id2path(child.file_id),
1863
child.file_id, "Child not deleted or reparented when "
1675
# If the two keys are the same, the value will be unchanged
1676
# as its always the file id.
1677
parent_id_basename_delta.append((old_key, new_key, new_value))
1865
1678
result.id_to_entry.apply_delta(id_to_entry_delta)
1866
1679
if parent_id_basename_delta:
1867
# Transform the parent_id_basename delta data into a linear delta
1868
# with only one record for a given key. Optimally this would allow
1869
# re-keying, but its simpler to just output that as a delete+add
1870
# to spend less time calculating the delta.
1872
for key, (old_key, value) in parent_id_basename_delta.iteritems():
1873
if value is not None:
1874
delta_list.append((old_key, key, value))
1876
delta_list.append((old_key, None, None))
1877
result.parent_id_basename_to_file_id.apply_delta(delta_list)
1878
parents.discard(('', None))
1879
for parent_path, parent in parents:
1680
result.parent_id_basename_to_file_id.apply_delta(parent_id_basename_delta)
1681
parents.discard(None)
1682
for parent in parents:
1881
1684
if result[parent].kind != 'directory':
1882
1685
raise errors.InconsistentDelta(result.id2path(parent), parent,
1918
1718
raise errors.BzrError('Duplicate key in inventory: %r\n%r'
1919
1719
% (key, bytes))
1920
1720
info[key] = value
1921
revision_id = intern(info['revision_id'])
1922
root_id = intern(info['root_id'])
1923
search_key_name = intern(info.get('search_key_name', 'plain'))
1924
parent_id_basename_to_file_id = intern(info.get(
1925
'parent_id_basename_to_file_id', None))
1926
if not parent_id_basename_to_file_id.startswith('sha1:'):
1927
raise ValueError('parent_id_basename_to_file_id should be a sha1'
1928
' key not %r' % (parent_id_basename_to_file_id,))
1721
revision_id = info['revision_id']
1722
root_id = info['root_id']
1723
search_key_name = info.get('search_key_name', 'plain')
1724
parent_id_basename_to_file_id = info.get(
1725
'parent_id_basename_to_file_id', None)
1929
1726
id_to_entry = info['id_to_entry']
1930
if not id_to_entry.startswith('sha1:'):
1931
raise ValueError('id_to_entry should be a sha1'
1932
' key not %r' % (id_to_entry,))
1934
1728
result = CHKInventory(search_key_name)
1935
1729
result.revision_id = revision_id
1938
1732
result._search_key_name)
1939
1733
if parent_id_basename_to_file_id is not None:
1940
1734
result.parent_id_basename_to_file_id = chk_map.CHKMap(
1941
chk_store, StaticTuple(parent_id_basename_to_file_id,),
1735
chk_store, (parent_id_basename_to_file_id,),
1942
1736
search_key_func=search_key_func)
1944
1738
result.parent_id_basename_to_file_id = None
1946
result.id_to_entry = chk_map.CHKMap(chk_store,
1947
StaticTuple(id_to_entry,),
1740
result.id_to_entry = chk_map.CHKMap(chk_store, (id_to_entry,),
1948
1741
search_key_func=search_key_func)
1949
1742
if (result.revision_id,) != expected_revision_id:
1950
1743
raise ValueError("Mismatched revision id and expected: %r, %r" %
2015
1807
return self._bytes_to_entry(
2016
self.id_to_entry.iteritems([StaticTuple(file_id,)]).next()[1])
1808
self.id_to_entry.iteritems([(file_id,)]).next()[1])
2017
1809
except StopIteration:
2018
1810
# really we're passing an inventory, not a tree...
2019
1811
raise errors.NoSuchId(self, file_id)
2021
def _getitems(self, file_ids):
2022
"""Similar to __getitem__, but lets you query for multiple.
2024
The returned order is undefined. And currently if an item doesn't
2025
exist, it isn't included in the output.
2029
for file_id in file_ids:
2030
entry = self._fileid_to_entry_cache.get(file_id, None)
2032
remaining.append(file_id)
2034
result.append(entry)
2035
file_keys = [StaticTuple(f,).intern() for f in remaining]
2036
for file_key, value in self.id_to_entry.iteritems(file_keys):
2037
entry = self._bytes_to_entry(value)
2038
result.append(entry)
2039
self._fileid_to_entry_cache[entry.file_id] = entry
2042
1813
def has_id(self, file_id):
2043
1814
# Perhaps have an explicit 'contains' method on CHKMap ?
2044
1815
if self._fileid_to_entry_cache.get(file_id, None) is not None:
2047
self.id_to_entry.iteritems([StaticTuple(file_id,)]))) == 1
1817
return len(list(self.id_to_entry.iteritems([(file_id,)]))) == 1
2049
1819
def is_root(self, file_id):
2050
1820
return file_id == self.root_id
2179
1949
delta.append((old_path, new_path, file_id, entry))
2182
def path2id(self, relpath):
1952
def path2id(self, name):
2183
1953
"""See CommonInventory.path2id()."""
2184
# TODO: perhaps support negative hits?
2185
result = self._path_to_fileid_cache.get(relpath, None)
2186
if result is not None:
2188
if isinstance(relpath, basestring):
2189
names = osutils.splitpath(relpath)
2192
current_id = self.root_id
2193
if current_id is None:
2195
parent_id_index = self.parent_id_basename_to_file_id
2197
for basename in names:
2198
if cur_path is None:
2201
cur_path = cur_path + '/' + basename
2202
basename_utf8 = basename.encode('utf8')
2203
file_id = self._path_to_fileid_cache.get(cur_path, None)
2205
key_filter = [StaticTuple(current_id, basename_utf8)]
2206
items = parent_id_index.iteritems(key_filter)
2207
for (parent_id, name_utf8), file_id in items:
2208
if parent_id != current_id or name_utf8 != basename_utf8:
2209
raise errors.BzrError("corrupt inventory lookup! "
2210
"%r %r %r %r" % (parent_id, current_id, name_utf8,
2215
self._path_to_fileid_cache[cur_path] = file_id
2216
current_id = file_id
1954
result = self._path_to_fileid_cache.get(name, None)
1956
result = CommonInventory.path2id(self, name)
1957
self._path_to_fileid_cache[name] = result
2219
1960
def to_lines(self):
2220
1961
"""Serialise the inventory to lines."""
2224
1965
lines.append('search_key_name: %s\n' % (self._search_key_name,))
2225
1966
lines.append("root_id: %s\n" % self.root_id)
2226
1967
lines.append('parent_id_basename_to_file_id: %s\n' %
2227
(self.parent_id_basename_to_file_id.key()[0],))
1968
self.parent_id_basename_to_file_id.key())
2228
1969
lines.append("revision_id: %s\n" % self.revision_id)
2229
lines.append("id_to_entry: %s\n" % (self.id_to_entry.key()[0],))
1970
lines.append("id_to_entry: %s\n" % self.id_to_entry.key())
2231
1972
lines.append("revision_id: %s\n" % self.revision_id)
2232
1973
lines.append("root_id: %s\n" % self.root_id)
2233
1974
if self.parent_id_basename_to_file_id is not None:
2234
1975
lines.append('parent_id_basename_to_file_id: %s\n' %
2235
(self.parent_id_basename_to_file_id.key()[0],))
2236
lines.append("id_to_entry: %s\n" % (self.id_to_entry.key()[0],))
1976
self.parent_id_basename_to_file_id.key())
1977
lines.append("id_to_entry: %s\n" % self.id_to_entry.key())
added
removed
bzrlib/inventory.py
