1
# Copyright (C) 2005-2011 Canonical Ltd
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
17
# FIXME: This refactoring of the workingtree code doesn't seem to keep
18
# the WorkingTree's copy of the inventory in sync with the branch. The
19
# branch modifies its working inventory when it does a commit to make
20
# missing files permanently removed.
22
# TODO: Maybe also keep the full path of the entry, and the children?
23
# But those depend on its position within a particular inventory, and
24
# it would be nice not to need to hold the backpointer here.
26
from __future__ import absolute_import
28
# This should really be an id randomly assigned when the tree is
29
# created, but it's not for now.
30
ROOT_ID = b"TREE_ROOT"
32
from ..lazy_import import lazy_import
33
lazy_import(globals(), """
44
from breezy.bzr import (
53
from ..sixish import (
60
from ..static_tuple import StaticTuple
63
class InventoryEntry(object):
64
"""Description of a versioned file.
66
An InventoryEntry has the following fields, which are also
67
present in the XML inventory-entry element:
72
(within the parent directory)
75
file_id of the parent directory, or ROOT_ID
78
the revision_id in which this variation of this file was
82
Indicates that this file should be executable on systems
86
sha-1 of the text of the file
89
size in bytes of the text of the file
91
(reading a version 4 tree created a text_id field.)
96
>>> i.add(InventoryDirectory('123', 'src', ROOT_ID))
97
InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None)
98
>>> i.add(InventoryFile('2323', 'hello.c', parent_id='123'))
99
InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None, revision=None)
100
>>> shouldbe = {0: '', 1: 'src', 2: 'src/hello.c'}
101
>>> for ix, j in enumerate(i.iter_entries()):
102
... print (j[0] == shouldbe[ix], j[1])
104
(True, InventoryDirectory('TREE_ROOT', u'', parent_id=None, revision=None))
105
(True, InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None))
106
(True, InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None, revision=None))
107
>>> i.add(InventoryFile('2324', 'bye.c', '123'))
108
InventoryFile('2324', 'bye.c', parent_id='123', sha1=None, len=None, revision=None)
109
>>> i.add(InventoryDirectory('2325', 'wibble', '123'))
110
InventoryDirectory('2325', 'wibble', parent_id='123', revision=None)
111
>>> i.path2id('src/wibble')
113
>>> i.add(InventoryFile('2326', 'wibble.c', '2325'))
114
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None, revision=None)
115
>>> i.get_entry('2326')
116
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None, revision=None)
117
>>> for path, entry in i.iter_entries():
126
>>> i.id2path(b'2326')
127
'src/wibble/wibble.c'
130
# Constants returned by describe_change()
132
# TODO: These should probably move to some kind of FileChangeDescription
133
# class; that's like what's inside a TreeDelta but we want to be able to
134
# generate them just for one file at a time.
136
MODIFIED_AND_RENAMED = 'modified and renamed'
138
__slots__ = ['file_id', 'revision', 'parent_id', 'name']
140
# Attributes that all InventoryEntry instances are expected to have, but
141
# that don't vary for all kinds of entry. (e.g. symlink_target is only
142
# relevant to InventoryLink, so there's no reason to make every
143
# InventoryFile instance allocate space to hold a value for it.)
144
# Attributes that only vary for files: executable, text_sha1, text_size,
150
# Attributes that only vary for symlinks: symlink_target
151
symlink_target = None
152
# Attributes that only vary for tree-references: reference_revision
153
reference_revision = None
156
def detect_changes(self, old_entry):
157
"""Return a (text_modified, meta_modified) from this to old_entry.
159
_read_tree_state must have been called on self and old_entry prior to
160
calling detect_changes.
164
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
165
output_to, reverse=False):
166
"""Perform a diff between two entries of the same kind."""
168
def parent_candidates(self, previous_inventories):
169
"""Find possible per-file graph parents.
171
This is currently defined by:
172
- Select the last changed revision in the parent inventory.
173
- Do deal with a short lived bug in bzr 0.8's development two entries
174
that have the same last changed but different 'x' bit settings are
177
# revision:ie mapping for each ie found in previous_inventories.
179
# identify candidate head revision ids.
180
for inv in previous_inventories:
182
ie = inv.get_entry(self.file_id)
183
except errors.NoSuchId:
186
if ie.revision in candidates:
187
# same revision value in two different inventories:
188
# correct possible inconsistencies:
189
# * there was a bug in revision updates with 'x' bit
192
if candidates[ie.revision].executable != ie.executable:
193
candidates[ie.revision].executable = False
194
ie.executable = False
195
except AttributeError:
198
# add this revision as a candidate.
199
candidates[ie.revision] = ie
203
"""Return true if the object this entry represents has textual data.
205
Note that textual data includes binary content.
207
Also note that all entries get weave files created for them.
208
This attribute is primarily used when upgrading from old trees that
209
did not have the weave index for all inventory entries.
213
def __init__(self, file_id, name, parent_id):
214
"""Create an InventoryEntry
216
The filename must be a single component, relative to the
217
parent directory; it cannot be a whole path or relative name.
219
>>> e = InventoryFile('123', 'hello.c', ROOT_ID)
224
>>> e = InventoryFile('123', 'src/hello.c', ROOT_ID)
225
Traceback (most recent call last):
226
InvalidEntryName: Invalid entry name: src/hello.c
229
raise errors.InvalidEntryName(name=name)
230
if not isinstance(file_id, bytes):
231
raise TypeError(file_id)
232
self.file_id = file_id
235
self.parent_id = parent_id
237
def kind_character(self):
238
"""Return a short kind indicator useful for appending to names."""
239
raise errors.BzrError('unknown kind %r' % self.kind)
241
known_kinds = ('file', 'directory', 'symlink')
244
def versionable_kind(kind):
245
return (kind in ('file', 'directory', 'symlink', 'tree-reference'))
247
def check(self, checker, rev_id, inv):
248
"""Check this inventory entry is intact.
250
This is a template method, override _check for kind specific
253
:param checker: Check object providing context for the checks;
254
can be used to find out what parts of the repository have already
256
:param rev_id: Revision id from which this InventoryEntry was loaded.
257
Not necessarily the last-changed revision for this file.
258
:param inv: Inventory from which the entry was loaded.
260
if self.parent_id is not None:
261
if not inv.has_id(self.parent_id):
262
raise errors.BzrCheckError(
263
'missing parent {%s} in inventory for revision {%s}' % (
264
self.parent_id, rev_id))
265
checker._add_entry_to_text_key_references(inv, self)
266
self._check(checker, rev_id)
268
def _check(self, checker, rev_id):
269
"""Check this inventory entry for kind specific errors."""
270
checker._report_items.append(
271
'unknown entry kind %r in revision {%s}' % (self.kind, rev_id))
274
"""Clone this inventory entry."""
275
raise NotImplementedError
278
def describe_change(old_entry, new_entry):
279
"""Describe the change between old_entry and this.
281
This smells of being an InterInventoryEntry situation, but as its
282
the first one, we're making it a static method for now.
284
An entry with a different parent, or different name is considered
285
to be renamed. Reparenting is an internal detail.
286
Note that renaming the parent does not trigger a rename for the
289
# TODO: Perhaps return an object rather than just a string
290
if old_entry is new_entry:
291
# also the case of both being None
293
elif old_entry is None:
295
elif new_entry is None:
297
if old_entry.kind != new_entry.kind:
299
text_modified, meta_modified = new_entry.detect_changes(old_entry)
300
if text_modified or meta_modified:
304
# TODO 20060511 (mbp, rbc) factor out 'detect_rename' here.
305
if old_entry.parent_id != new_entry.parent_id:
307
elif old_entry.name != new_entry.name:
311
if renamed and not modified:
312
return InventoryEntry.RENAMED
313
if modified and not renamed:
315
if modified and renamed:
316
return InventoryEntry.MODIFIED_AND_RENAMED
320
return ("%s(%r, %r, parent_id=%r, revision=%r)"
321
% (self.__class__.__name__,
327
def __eq__(self, other):
329
# For the case when objects are cached
331
if not isinstance(other, InventoryEntry):
332
return NotImplemented
334
return ((self.file_id == other.file_id)
335
and (self.name == other.name)
336
and (other.symlink_target == self.symlink_target)
337
and (self.text_sha1 == other.text_sha1)
338
and (self.text_size == other.text_size)
339
and (self.text_id == other.text_id)
340
and (self.parent_id == other.parent_id)
341
and (self.kind == other.kind)
342
and (self.revision == other.revision)
343
and (self.executable == other.executable)
344
and (self.reference_revision == other.reference_revision)
347
def __ne__(self, other):
348
return not (self == other)
351
raise ValueError('not hashable')
353
def _unchanged(self, previous_ie):
354
"""Has this entry changed relative to previous_ie.
356
This method should be overridden in child classes.
359
# different inv parent
360
if previous_ie.parent_id != self.parent_id:
363
elif previous_ie.name != self.name:
365
elif previous_ie.kind != self.kind:
369
def _read_tree_state(self, path, work_tree):
370
"""Populate fields in the inventory entry from the given tree.
372
Note that this should be modified to be a noop on virtual trees
373
as all entries created there are prepopulated.
375
# TODO: Rather than running this manually, we should check the
376
# working sha1 and other expensive properties when they're
377
# first requested, or preload them if they're already known
378
pass # nothing to do by default
380
def _forget_tree_state(self):
384
class InventoryDirectory(InventoryEntry):
385
"""A directory in an inventory."""
387
__slots__ = ['children']
391
def _check(self, checker, rev_id):
392
"""See InventoryEntry._check"""
393
# In non rich root repositories we do not expect a file graph for the
395
if self.name == '' and not checker.rich_roots:
397
# Directories are stored as an empty file, but the file should exist
398
# to provide a per-fileid log. The hash of every directory content is
399
# "da..." below (the sha1sum of '').
400
checker.add_pending_item(rev_id,
401
('texts', self.file_id, self.revision), b'text',
402
b'da39a3ee5e6b4b0d3255bfef95601890afd80709')
405
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
406
other.revision = self.revision
407
# note that children are *not* copied; they're pulled across when
411
def __init__(self, file_id, name, parent_id):
412
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
415
def sorted_children(self):
416
return sorted(viewitems(self.children))
418
def kind_character(self):
419
"""See InventoryEntry.kind_character."""
423
class InventoryFile(InventoryEntry):
424
"""A file in an inventory."""
426
__slots__ = ['text_sha1', 'text_size', 'text_id', 'executable']
430
def __init__(self, file_id, name, parent_id):
431
super(InventoryFile, self).__init__(file_id, name, parent_id)
432
self.text_sha1 = None
433
self.text_size = None
435
self.executable = False
437
def _check(self, checker, tree_revision_id):
438
"""See InventoryEntry._check"""
439
# TODO: check size too.
440
checker.add_pending_item(tree_revision_id,
441
('texts', self.file_id, self.revision), b'text',
443
if self.text_size is None:
444
checker._report_items.append(
445
'fileid {%s} in {%s} has None for text_size' % (self.file_id,
449
other = InventoryFile(self.file_id, self.name, self.parent_id)
450
other.executable = self.executable
451
other.text_id = self.text_id
452
other.text_sha1 = self.text_sha1
453
other.text_size = self.text_size
454
other.revision = self.revision
457
def detect_changes(self, old_entry):
458
"""See InventoryEntry.detect_changes."""
459
text_modified = (self.text_sha1 != old_entry.text_sha1)
460
meta_modified = (self.executable != old_entry.executable)
461
return text_modified, meta_modified
463
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
464
output_to, reverse=False):
465
"""See InventoryEntry._diff."""
466
from breezy.diff import DiffText
467
from_file_id = self.file_id
469
to_file_id = to_entry.file_id
470
to_path = to_tree.id2path(to_file_id)
474
if from_file_id is not None:
475
from_path = tree.id2path(from_file_id)
479
to_file_id, from_file_id = from_file_id, to_file_id
480
tree, to_tree = to_tree, tree
481
from_label, to_label = to_label, from_label
482
differ = DiffText(tree, to_tree, output_to, 'utf-8', '', '',
484
return differ.diff_text(from_path, to_path, from_label, to_label,
485
from_file_id, to_file_id)
488
"""See InventoryEntry.has_text."""
491
def kind_character(self):
492
"""See InventoryEntry.kind_character."""
495
def _read_tree_state(self, path, work_tree):
496
"""See InventoryEntry._read_tree_state."""
497
self.text_sha1 = work_tree.get_file_sha1(path, self.file_id)
498
# FIXME: 20050930 probe for the text size when getting sha1
499
# in _read_tree_state
500
self.executable = work_tree.is_executable(path, self.file_id)
503
return ("%s(%r, %r, parent_id=%r, sha1=%r, len=%s, revision=%s)"
504
% (self.__class__.__name__,
512
def _forget_tree_state(self):
513
self.text_sha1 = None
515
def _unchanged(self, previous_ie):
516
"""See InventoryEntry._unchanged."""
517
compatible = super(InventoryFile, self)._unchanged(previous_ie)
518
if self.text_sha1 != previous_ie.text_sha1:
521
# FIXME: 20050930 probe for the text size when getting sha1
522
# in _read_tree_state
523
self.text_size = previous_ie.text_size
524
if self.executable != previous_ie.executable:
529
class InventoryLink(InventoryEntry):
530
"""A file in an inventory."""
532
__slots__ = ['symlink_target']
536
def __init__(self, file_id, name, parent_id):
537
super(InventoryLink, self).__init__(file_id, name, parent_id)
538
self.symlink_target = None
540
def _check(self, checker, tree_revision_id):
541
"""See InventoryEntry._check"""
542
if self.symlink_target is None:
543
checker._report_items.append(
544
'symlink {%s} has no target in revision {%s}'
545
% (self.file_id, tree_revision_id))
546
# Symlinks are stored as ''
547
checker.add_pending_item(tree_revision_id,
548
('texts', self.file_id, self.revision), b'text',
549
b'da39a3ee5e6b4b0d3255bfef95601890afd80709')
552
other = InventoryLink(self.file_id, self.name, self.parent_id)
553
other.symlink_target = self.symlink_target
554
other.revision = self.revision
557
def detect_changes(self, old_entry):
558
"""See InventoryEntry.detect_changes."""
559
# FIXME: which _modified field should we use ? RBC 20051003
560
text_modified = (self.symlink_target != old_entry.symlink_target)
562
trace.mutter(" symlink target changed")
563
meta_modified = False
564
return text_modified, meta_modified
566
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
567
output_to, reverse=False):
568
"""See InventoryEntry._diff."""
569
from breezy.diff import DiffSymlink
570
old_target = self.symlink_target
571
if to_entry is not None:
572
new_target = to_entry.symlink_target
581
new_target, old_target = old_target, new_target
582
differ = DiffSymlink(old_tree, new_tree, output_to)
583
return differ.diff_symlink(old_target, new_target)
585
def kind_character(self):
586
"""See InventoryEntry.kind_character."""
589
def _read_tree_state(self, path, work_tree):
590
"""See InventoryEntry._read_tree_state."""
591
self.symlink_target = work_tree.get_symlink_target(
592
work_tree.id2path(self.file_id), self.file_id)
594
def _forget_tree_state(self):
595
self.symlink_target = None
597
def _unchanged(self, previous_ie):
598
"""See InventoryEntry._unchanged."""
599
compatible = super(InventoryLink, self)._unchanged(previous_ie)
600
if self.symlink_target != previous_ie.symlink_target:
605
class TreeReference(InventoryEntry):
607
__slots__ = ['reference_revision']
609
kind = 'tree-reference'
611
def __init__(self, file_id, name, parent_id, revision=None,
612
reference_revision=None):
613
InventoryEntry.__init__(self, file_id, name, parent_id)
614
self.revision = revision
615
self.reference_revision = reference_revision
618
return TreeReference(self.file_id, self.name, self.parent_id,
619
self.revision, self.reference_revision)
621
def _read_tree_state(self, path, work_tree):
622
"""Populate fields in the inventory entry from the given tree.
624
self.reference_revision = work_tree.get_reference_revision(
627
def _forget_tree_state(self):
628
self.reference_revision = None
630
def _unchanged(self, previous_ie):
631
"""See InventoryEntry._unchanged."""
632
compatible = super(TreeReference, self)._unchanged(previous_ie)
633
if self.reference_revision != previous_ie.reference_revision:
638
class CommonInventory(object):
639
"""Basic inventory logic, defined in terms of primitives like has_id.
641
An inventory is the metadata about the contents of a tree.
643
This is broadly a map from file_id to entries such as directories, files,
644
symlinks and tree references. Each entry maintains its own metadata like
645
SHA1 and length for files, or children for a directory.
647
Entries can be looked up either by path or by file_id.
649
InventoryEntry objects must not be modified after they are
650
inserted, other than through the Inventory API.
653
def has_filename(self, filename):
654
return bool(self.path2id(filename))
656
def id2path(self, file_id):
657
"""Return as a string the path to file_id.
660
>>> e = i.add(InventoryDirectory(b'src-id', 'src', ROOT_ID))
661
>>> e = i.add(InventoryFile(b'foo-id', 'foo.c', parent_id='src-id'))
662
>>> print i.id2path(b'foo-id')
665
:raises NoSuchId: If file_id is not present in the inventory.
667
# get all names, skipping root
668
return '/'.join(reversed(
669
[parent.name for parent in
670
self._iter_file_id_parents(file_id)][:-1]))
672
def iter_entries(self, from_dir=None, recursive=True):
673
"""Return (path, entry) pairs, in order by name.
675
:param from_dir: if None, start from the root,
676
otherwise start from this directory (either file-id or entry)
677
:param recursive: recurse into directories or not
680
if self.root is None:
684
elif isinstance(from_dir, bytes):
685
from_dir = self.get_entry(from_dir)
687
# unrolling the recursive called changed the time from
688
# 440ms/663ms (inline/total) to 116ms/116ms
689
children = sorted(viewitems(from_dir.children))
691
for name, ie in children:
694
children = collections.deque(children)
695
stack = [(u'', children)]
697
from_dir_relpath, children = stack[-1]
700
name, ie = children.popleft()
702
# we know that from_dir_relpath never ends in a slash
703
# and 'f' doesn't begin with one, we can do a string op, rather
704
# than the checks of pathjoin(), though this means that all paths
706
path = from_dir_relpath + '/' + name
710
if ie.kind != 'directory':
713
# But do this child first
714
new_children = sorted(viewitems(ie.children))
715
new_children = collections.deque(new_children)
716
stack.append((path, new_children))
717
# Break out of inner loop, so that we start outer loop with child
720
# if we finished all children, pop it off the stack
723
def _preload_cache(self):
724
"""Populate any caches, we are about to access all items.
726
The default implementation does nothing, because CommonInventory doesn't
731
def iter_entries_by_dir(self, from_dir=None, specific_file_ids=None):
732
"""Iterate over the entries in a directory first order.
734
This returns all entries for a directory before returning
735
the entries for children of a directory. This is not
736
lexicographically sorted order, and is a hybrid between
737
depth-first and breadth-first.
739
:return: This yields (path, entry) pairs
741
if specific_file_ids and not isinstance(specific_file_ids, set):
742
specific_file_ids = set(specific_file_ids)
743
# TODO? Perhaps this should return the from_dir so that the root is
744
# yielded? or maybe an option?
745
if from_dir is None and specific_file_ids is None:
746
# They are iterating from the root, and have not specified any
747
# specific entries to look at. All current callers fully consume the
748
# iterator, so we can safely assume we are accessing all entries
749
self._preload_cache()
751
if self.root is None:
753
# Optimize a common case
754
if (specific_file_ids is not None and
755
len(specific_file_ids) == 1):
756
file_id = list(specific_file_ids)[0]
757
if file_id is not None:
759
path = self.id2path(file_id)
760
except errors.NoSuchId:
763
yield path, self.get_entry(file_id)
766
if (specific_file_ids is None or
767
self.root.file_id in specific_file_ids):
769
elif isinstance(from_dir, bytes):
770
from_dir = self.get_entry(from_dir)
772
raise TypeError(from_dir)
774
if specific_file_ids is not None:
775
# TODO: jam 20070302 This could really be done as a loop rather
776
# than a bunch of recursive calls.
779
def add_ancestors(file_id):
780
if not byid.has_id(file_id):
782
parent_id = byid.get_entry(file_id).parent_id
783
if parent_id is None:
785
if parent_id not in parents:
786
parents.add(parent_id)
787
add_ancestors(parent_id)
788
for file_id in specific_file_ids:
789
add_ancestors(file_id)
793
stack = [(u'', from_dir)]
795
cur_relpath, cur_dir = stack.pop()
798
for child_name, child_ie in sorted(viewitems(cur_dir.children)):
800
child_relpath = cur_relpath + child_name
802
if (specific_file_ids is None or
803
child_ie.file_id in specific_file_ids):
804
yield child_relpath, child_ie
806
if child_ie.kind == 'directory':
807
if parents is None or child_ie.file_id in parents:
808
child_dirs.append((child_relpath+'/', child_ie))
809
stack.extend(reversed(child_dirs))
811
def _make_delta(self, old):
812
"""Make an inventory delta from two inventories."""
813
old_ids = set(old.iter_all_ids())
814
new_ids = set(self.iter_all_ids())
815
adds = new_ids - old_ids
816
deletes = old_ids - new_ids
817
common = old_ids.intersection(new_ids)
819
for file_id in deletes:
820
delta.append((old.id2path(file_id), None, file_id, None))
822
delta.append((None, self.id2path(file_id), file_id, self.get_entry(file_id)))
823
for file_id in common:
824
if old.get_entry(file_id) != self.get_entry(file_id):
825
delta.append((old.id2path(file_id), self.id2path(file_id),
826
file_id, self.get_entry(file_id)))
829
def make_entry(self, kind, name, parent_id, file_id=None):
830
"""Simple thunk to breezy.bzr.inventory.make_entry."""
831
return make_entry(kind, name, parent_id, file_id)
834
"""Return list of (path, ie) for all entries except the root.
836
This may be faster than iter_entries.
839
def descend(dir_ie, dir_path):
840
kids = sorted(viewitems(dir_ie.children))
841
for name, ie in kids:
842
child_path = osutils.pathjoin(dir_path, name)
843
accum.append((child_path, ie))
844
if ie.kind == 'directory':
845
descend(ie, child_path)
847
if self.root is not None:
848
descend(self.root, u'')
851
def get_entry_by_path(self, relpath):
852
"""Return an inventory entry by path.
854
:param relpath: may be either a list of path components, or a single
855
string, in which case it is automatically split.
857
This returns the entry of the last component in the path,
858
which may be either a file or a directory.
860
Returns None IFF the path is not found.
862
if isinstance(relpath, (str, text_type)):
863
names = osutils.splitpath(relpath)
869
except errors.NoSuchId:
870
# root doesn't exist yet so nothing else can
876
children = getattr(parent, 'children', None)
886
def path2id(self, relpath):
887
"""Walk down through directories to return entry of last component.
889
:param relpath: may be either a list of path components, or a single
890
string, in which case it is automatically split.
892
This returns the entry of the last component in the path,
893
which may be either a file or a directory.
895
Returns None IFF the path is not found.
897
ie = self.get_entry_by_path(relpath)
902
def filter(self, specific_fileids):
903
"""Get an inventory view filtered against a set of file-ids.
905
Children of directories and parents are included.
907
The result may or may not reference the underlying inventory
908
so it should be treated as immutable.
910
interesting_parents = set()
911
for fileid in specific_fileids:
913
interesting_parents.update(self.get_idpath(fileid))
914
except errors.NoSuchId:
915
# This fileid is not in the inventory - that's ok
917
entries = self.iter_entries()
918
if self.root is None:
919
return Inventory(root_id=None)
920
other = Inventory(next(entries)[1].file_id)
921
other.root.revision = self.root.revision
922
other.revision_id = self.revision_id
923
directories_to_expand = set()
924
for path, entry in entries:
925
file_id = entry.file_id
926
if (file_id in specific_fileids
927
or entry.parent_id in directories_to_expand):
928
if entry.kind == 'directory':
929
directories_to_expand.add(file_id)
930
elif file_id not in interesting_parents:
932
other.add(entry.copy())
935
def get_idpath(self, file_id):
936
"""Return a list of file_ids for the path to an entry.
938
The list contains one element for each directory followed by
939
the id of the file itself. So the length of the returned list
940
is equal to the depth of the file in the tree, counting the
941
root directory as depth 1.
944
for parent in self._iter_file_id_parents(file_id):
945
p.insert(0, parent.file_id)
949
class Inventory(CommonInventory):
950
"""Mutable dict based in-memory inventory.
952
We never store the full path to a file, because renaming a directory
953
implicitly moves all of its contents. This class internally maintains a
954
lookup tree that allows the children under a directory to be
957
>>> inv = Inventory()
958
>>> inv.add(InventoryFile(b'123-123', 'hello.c', ROOT_ID))
959
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
960
>>> inv.get_entry(b'123-123').name
963
Id's may be looked up from paths:
965
>>> inv.path2id('hello.c')
967
>>> inv.has_id(b'123-123')
970
There are iterators over the contents:
972
>>> [entry[0] for entry in inv.iter_entries()]
976
def __init__(self, root_id=ROOT_ID, revision_id=None):
977
"""Create or read an inventory.
979
If a working directory is specified, the inventory is read
980
from there. If the file is specified, read from that. If not,
981
the inventory is created empty.
983
The inventory is created with a default root directory, with
986
if root_id is not None:
987
self._set_root(InventoryDirectory(root_id, u'', None))
991
self.revision_id = revision_id
994
# More than one page of ouput is not useful anymore to debug
997
contents = repr(self._byid)
998
if len(contents) > max_len:
999
contents = contents[:(max_len-len(closing))] + closing
1000
return "<Inventory object at %x, contents=%r>" % (id(self), contents)
1002
def apply_delta(self, delta):
1003
"""Apply a delta to this inventory.
1005
See the inventory developers documentation for the theory behind
1008
If delta application fails the inventory is left in an indeterminate
1009
state and must not be used.
1011
:param delta: A list of changes to apply. After all the changes are
1012
applied the final inventory must be internally consistent, but it
1013
is ok to supply changes which, if only half-applied would have an
1014
invalid result - such as supplying two changes which rename two
1015
files, 'A' and 'B' with each other : [('A', 'B', b'A-id', a_entry),
1016
('B', 'A', b'B-id', b_entry)].
1018
Each change is a tuple, of the form (old_path, new_path, file_id,
1021
When new_path is None, the change indicates the removal of an entry
1022
from the inventory and new_entry will be ignored (using None is
1023
appropriate). If new_path is not None, then new_entry must be an
1024
InventoryEntry instance, which will be incorporated into the
1025
inventory (and replace any existing entry with the same file id).
1027
When old_path is None, the change indicates the addition of
1028
a new entry to the inventory.
1030
When neither new_path nor old_path are None, the change is a
1031
modification to an entry, such as a rename, reparent, kind change
1034
The children attribute of new_entry is ignored. This is because
1035
this method preserves children automatically across alterations to
1036
the parent of the children, and cases where the parent id of a
1037
child is changing require the child to be passed in as a separate
1038
change regardless. E.g. in the recursive deletion of a directory -
1039
the directory's children must be included in the delta, or the
1040
final inventory will be invalid.
1042
Note that a file_id must only appear once within a given delta.
1043
An AssertionError is raised otherwise.
1045
# Check that the delta is legal. It would be nice if this could be
1046
# done within the loops below but it's safer to validate the delta
1047
# before starting to mutate the inventory, as there isn't a rollback
1049
list(_check_delta_unique_ids(_check_delta_unique_new_paths(
1050
_check_delta_unique_old_paths(_check_delta_ids_match_entry(
1051
_check_delta_ids_are_valid(
1052
_check_delta_new_path_entry_both_or_None(
1056
# Remove all affected items which were in the original inventory,
1057
# starting with the longest paths, thus ensuring parents are examined
1058
# after their children, which means that everything we examine has no
1059
# modified children remaining by the time we examine it.
1060
for old_path, file_id in sorted(((op, f) for op, np, f, e in delta
1061
if op is not None), reverse=True):
1062
# Preserve unaltered children of file_id for later reinsertion.
1063
file_id_children = getattr(self.get_entry(file_id), 'children', {})
1064
if len(file_id_children):
1065
children[file_id] = file_id_children
1066
if self.id2path(file_id) != old_path:
1067
raise errors.InconsistentDelta(old_path, file_id,
1068
"Entry was at wrong other path %r." % self.id2path(file_id))
1069
# Remove file_id and the unaltered children. If file_id is not
1070
# being deleted it will be reinserted back later.
1071
self.remove_recursive_id(file_id)
1072
# Insert all affected which should be in the new inventory, reattaching
1073
# their children if they had any. This is done from shortest path to
1074
# longest, ensuring that items which were modified and whose parents in
1075
# the resulting inventory were also modified, are inserted after their
1077
for new_path, f, new_entry in sorted((np, f, e) for op, np, f, e in
1078
delta if np is not None):
1079
if new_entry.kind == 'directory':
1080
# Pop the child which to allow detection of children whose
1081
# parents were deleted and which were not reattached to a new
1083
replacement = InventoryDirectory(new_entry.file_id,
1084
new_entry.name, new_entry.parent_id)
1085
replacement.revision = new_entry.revision
1086
replacement.children = children.pop(replacement.file_id, {})
1087
new_entry = replacement
1090
except errors.DuplicateFileId:
1091
raise errors.InconsistentDelta(new_path, new_entry.file_id,
1092
"New id is already present in target.")
1093
except AttributeError:
1094
raise errors.InconsistentDelta(new_path, new_entry.file_id,
1095
"Parent is not a directory.")
1096
if self.id2path(new_entry.file_id) != new_path:
1097
raise errors.InconsistentDelta(new_path, new_entry.file_id,
1098
"New path is not consistent with parent path.")
1100
# Get the parent id that was deleted
1101
parent_id, children = children.popitem()
1102
raise errors.InconsistentDelta("<deleted>", parent_id,
1103
"The file id was deleted but its children were not deleted.")
1105
def create_by_apply_delta(self, inventory_delta, new_revision_id,
1106
propagate_caches=False):
1107
"""See CHKInventory.create_by_apply_delta()"""
1108
new_inv = self.copy()
1109
new_inv.apply_delta(inventory_delta)
1110
new_inv.revision_id = new_revision_id
1113
def _set_root(self, ie):
1115
self._byid = {self.root.file_id: self.root}
1118
# TODO: jam 20051218 Should copy also copy the revision_id?
1119
entries = self.iter_entries()
1120
if self.root is None:
1121
return Inventory(root_id=None)
1122
other = Inventory(next(entries)[1].file_id)
1123
other.root.revision = self.root.revision
1124
# copy recursively so we know directories will be added before
1125
# their children. There are more efficient ways than this...
1126
for path, entry in entries:
1127
other.add(entry.copy())
1130
def iter_all_ids(self):
1131
"""Iterate over all file-ids."""
1132
return iter(self._byid)
1134
def iter_just_entries(self):
1135
"""Iterate over all entries.
1137
Unlike iter_entries(), just the entries are returned (not (path, ie))
1138
and the order of entries is undefined.
1140
XXX: We may not want to merge this into bzr.dev.
1142
if self.root is None:
1144
return iter(viewvalues(self._byid))
1147
"""Returns number of entries."""
1148
return len(self._byid)
1150
def get_entry(self, file_id):
1151
"""Return the entry for given file_id.
1153
>>> inv = Inventory()
1154
>>> inv.add(InventoryFile(b'123123', 'hello.c', ROOT_ID))
1155
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
1156
>>> inv.get_entry(b'123123').name
1159
if not isinstance(file_id, bytes):
1160
raise TypeError(file_id)
1162
return self._byid[file_id]
1164
# really we're passing an inventory, not a tree...
1165
raise errors.NoSuchId(self, file_id)
1167
def get_file_kind(self, file_id):
1168
return self._byid[file_id].kind
1170
def get_child(self, parent_id, filename):
1171
return self.get_entry(parent_id).children.get(filename)
1173
def _add_child(self, entry):
1174
"""Add an entry to the inventory, without adding it to its parent"""
1175
if entry.file_id in self._byid:
1176
raise errors.BzrError(
1177
"inventory already contains entry with id {%s}" %
1179
self._byid[entry.file_id] = entry
1180
children = getattr(entry, 'children', {})
1181
if children is not None:
1182
for child in viewvalues(children):
1183
self._add_child(child)
1186
def add(self, entry):
1187
"""Add entry to inventory.
1191
if entry.file_id in self._byid:
1192
raise errors.DuplicateFileId(entry.file_id,
1193
self._byid[entry.file_id])
1194
if entry.parent_id is None:
1198
parent = self._byid[entry.parent_id]
1200
raise errors.InconsistentDelta("<unknown>", entry.parent_id,
1201
"Parent not in inventory.")
1202
if entry.name in parent.children:
1203
raise errors.InconsistentDelta(
1204
self.id2path(parent.children[entry.name].file_id),
1206
"Path already versioned")
1207
parent.children[entry.name] = entry
1208
return self._add_child(entry)
1210
def add_path(self, relpath, kind, file_id=None, parent_id=None):
1211
"""Add entry from a path.
1213
The immediate parent must already be versioned.
1215
Returns the new entry object."""
1217
parts = osutils.splitpath(relpath)
1221
file_id = generate_ids.gen_root_id()
1222
self.root = InventoryDirectory(file_id, '', None)
1223
self._byid = {self.root.file_id: self.root}
1226
parent_path = parts[:-1]
1227
parent_id = self.path2id(parent_path)
1228
if parent_id is None:
1229
raise errors.NotVersionedError(path=parent_path)
1230
ie = make_entry(kind, parts[-1], parent_id, file_id)
1233
def delete(self, file_id):
1234
"""Remove entry by id.
1236
>>> inv = Inventory()
1237
>>> inv.add(InventoryFile(b'123', 'foo.c', ROOT_ID))
1238
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
1239
>>> inv.has_id(b'123')
1241
>>> inv.delete(b'123')
1242
>>> inv.has_id(b'123')
1245
ie = self.get_entry(file_id)
1246
del self._byid[file_id]
1247
if ie.parent_id is not None:
1248
del self.get_entry(ie.parent_id).children[ie.name]
1250
def __eq__(self, other):
1251
"""Compare two sets by comparing their contents.
1253
>>> i1 = Inventory()
1254
>>> i2 = Inventory()
1257
>>> i1.add(InventoryFile(b'123', 'foo', ROOT_ID))
1258
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
1261
>>> i2.add(InventoryFile(b'123', 'foo', ROOT_ID))
1262
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
1266
if not isinstance(other, Inventory):
1267
return NotImplemented
1269
return self._byid == other._byid
1271
def __ne__(self, other):
1272
return not self.__eq__(other)
1275
raise ValueError('not hashable')
1277
def _iter_file_id_parents(self, file_id):
1278
"""Yield the parents of file_id up to the root."""
1279
while file_id is not None:
1281
ie = self._byid[file_id]
1283
raise errors.NoSuchId(tree=None, file_id=file_id)
1285
file_id = ie.parent_id
1287
def has_id(self, file_id):
1288
return (file_id in self._byid)
1290
def _make_delta(self, old):
1291
"""Make an inventory delta from two inventories."""
1292
old_getter = old.get_entry
1293
new_getter = self.get_entry
1294
old_ids = set(old.iter_all_ids())
1295
new_ids = set(self.iter_all_ids())
1296
adds = new_ids - old_ids
1297
deletes = old_ids - new_ids
1298
if not adds and not deletes:
1301
common = old_ids.intersection(new_ids)
1303
for file_id in deletes:
1304
delta.append((old.id2path(file_id), None, file_id, None))
1305
for file_id in adds:
1306
delta.append((None, self.id2path(file_id), file_id, self.get_entry(file_id)))
1307
for file_id in common:
1308
new_ie = new_getter(file_id)
1309
old_ie = old_getter(file_id)
1310
# If xml_serializer returns the cached InventoryEntries (rather
1311
# than always doing .copy()), inlining the 'is' check saves 2.7M
1312
# calls to __eq__. Under lsprof this saves 20s => 6s.
1313
# It is a minor improvement without lsprof.
1314
if old_ie is new_ie or old_ie == new_ie:
1317
delta.append((old.id2path(file_id), self.id2path(file_id),
1321
def remove_recursive_id(self, file_id):
1322
"""Remove file_id, and children, from the inventory.
1324
:param file_id: A file_id to remove.
1326
to_find_delete = [self._byid[file_id]]
1328
while to_find_delete:
1329
ie = to_find_delete.pop()
1330
to_delete.append(ie.file_id)
1331
if ie.kind == 'directory':
1332
to_find_delete.extend(viewvalues(ie.children))
1333
for file_id in reversed(to_delete):
1334
ie = self.get_entry(file_id)
1335
del self._byid[file_id]
1336
if ie.parent_id is not None:
1337
del self.get_entry(ie.parent_id).children[ie.name]
1341
def rename(self, file_id, new_parent_id, new_name):
1342
"""Move a file within the inventory.
1344
This can change either the name, or the parent, or both.
1346
This does not move the working file.
1348
new_name = ensure_normalized_name(new_name)
1349
if not is_valid_name(new_name):
1350
raise errors.BzrError("not an acceptable filename: %r" % new_name)
1352
new_parent = self._byid[new_parent_id]
1353
if new_name in new_parent.children:
1354
raise errors.BzrError("%r already exists in %r" %
1355
(new_name, self.id2path(new_parent_id)))
1357
new_parent_idpath = self.get_idpath(new_parent_id)
1358
if file_id in new_parent_idpath:
1359
raise errors.BzrError(
1360
"cannot move directory %r into a subdirectory of itself, %r"
1361
% (self.id2path(file_id), self.id2path(new_parent_id)))
1363
file_ie = self._byid[file_id]
1364
old_parent = self._byid[file_ie.parent_id]
1366
# TODO: Don't leave things messed up if this fails
1368
del old_parent.children[file_ie.name]
1369
new_parent.children[new_name] = file_ie
1371
file_ie.name = new_name
1372
file_ie.parent_id = new_parent_id
1374
def is_root(self, file_id):
1375
return self.root is not None and file_id == self.root.file_id
1378
class CHKInventory(CommonInventory):
1379
"""An inventory persisted in a CHK store.
1381
By design, a CHKInventory is immutable so many of the methods
1382
supported by Inventory - add, rename, apply_delta, etc - are *not*
1383
supported. To create a new CHKInventory, use create_by_apply_delta()
1384
or from_inventory(), say.
1386
Internally, a CHKInventory has one or two CHKMaps:
1388
* id_to_entry - a map from (file_id,) => InventoryEntry as bytes
1389
* parent_id_basename_to_file_id - a map from (parent_id, basename_utf8)
1392
The second map is optional and not present in early CHkRepository's.
1394
No caching is performed: every method call or item access will perform
1395
requests to the storage layer. As such, keep references to objects you
1399
def __init__(self, search_key_name):
1400
CommonInventory.__init__(self)
1401
self._fileid_to_entry_cache = {}
1402
self._fully_cached = False
1403
self._path_to_fileid_cache = {}
1404
self._search_key_name = search_key_name
1407
def __eq__(self, other):
1408
"""Compare two sets by comparing their contents."""
1409
if not isinstance(other, CHKInventory):
1410
return NotImplemented
1412
this_key = self.id_to_entry.key()
1413
other_key = other.id_to_entry.key()
1414
this_pid_key = self.parent_id_basename_to_file_id.key()
1415
other_pid_key = other.parent_id_basename_to_file_id.key()
1416
if None in (this_key, this_pid_key, other_key, other_pid_key):
1418
return this_key == other_key and this_pid_key == other_pid_key
1420
def _entry_to_bytes(self, entry):
1421
"""Serialise entry as a single bytestring.
1423
:param Entry: An inventory entry.
1424
:return: A bytestring for the entry.
1427
ENTRY ::= FILE | DIR | SYMLINK | TREE
1428
FILE ::= "file: " COMMON SEP SHA SEP SIZE SEP EXECUTABLE
1429
DIR ::= "dir: " COMMON
1430
SYMLINK ::= "symlink: " COMMON SEP TARGET_UTF8
1431
TREE ::= "tree: " COMMON REFERENCE_REVISION
1432
COMMON ::= FILE_ID SEP PARENT_ID SEP NAME_UTF8 SEP REVISION
1435
if entry.parent_id is not None:
1436
parent_str = entry.parent_id
1439
name_str = entry.name.encode("utf8")
1440
if entry.kind == 'file':
1441
if entry.executable:
1445
return b"file: %s\n%s\n%s\n%s\n%s\n%d\n%s" % (
1446
entry.file_id, parent_str, name_str, entry.revision,
1447
entry.text_sha1, entry.text_size, exec_str)
1448
elif entry.kind == 'directory':
1449
return b"dir: %s\n%s\n%s\n%s" % (
1450
entry.file_id, parent_str, name_str, entry.revision)
1451
elif entry.kind == 'symlink':
1452
return b"symlink: %s\n%s\n%s\n%s\n%s" % (
1453
entry.file_id, parent_str, name_str, entry.revision,
1454
entry.symlink_target.encode("utf8"))
1455
elif entry.kind == 'tree-reference':
1456
return b"tree: %s\n%s\n%s\n%s\n%s" % (
1457
entry.file_id, parent_str, name_str, entry.revision,
1458
entry.reference_revision)
1460
raise ValueError("unknown kind %r" % entry.kind)
1462
def _expand_fileids_to_parents_and_children(self, file_ids):
1463
"""Give a more wholistic view starting with the given file_ids.
1465
For any file_id which maps to a directory, we will include all children
1466
of that directory. We will also include all directories which are
1467
parents of the given file_ids, but we will not include their children.
1474
fringle # fringle-id
1478
if given [foo-id] we will include
1479
TREE_ROOT as interesting parents
1481
foo-id, baz-id, frob-id, fringle-id
1485
# TODO: Pre-pass over the list of fileids to see if anything is already
1486
# deserialized in self._fileid_to_entry_cache
1488
directories_to_expand = set()
1489
children_of_parent_id = {}
1490
# It is okay if some of the fileids are missing
1491
for entry in self._getitems(file_ids):
1492
if entry.kind == 'directory':
1493
directories_to_expand.add(entry.file_id)
1494
interesting.add(entry.parent_id)
1495
children_of_parent_id.setdefault(entry.parent_id, set()
1496
).add(entry.file_id)
1498
# Now, interesting has all of the direct parents, but not the
1499
# parents of those parents. It also may have some duplicates with
1501
remaining_parents = interesting.difference(file_ids)
1502
# When we hit the TREE_ROOT, we'll get an interesting parent of None,
1503
# but we don't actually want to recurse into that
1504
interesting.add(None) # this will auto-filter it in the loop
1505
remaining_parents.discard(None)
1506
while remaining_parents:
1507
next_parents = set()
1508
for entry in self._getitems(remaining_parents):
1509
next_parents.add(entry.parent_id)
1510
children_of_parent_id.setdefault(entry.parent_id, set()
1511
).add(entry.file_id)
1512
# Remove any search tips we've already processed
1513
remaining_parents = next_parents.difference(interesting)
1514
interesting.update(remaining_parents)
1515
# We should probably also .difference(directories_to_expand)
1516
interesting.update(file_ids)
1517
interesting.discard(None)
1518
while directories_to_expand:
1519
# Expand directories by looking in the
1520
# parent_id_basename_to_file_id map
1521
keys = [StaticTuple(f,).intern() for f in directories_to_expand]
1522
directories_to_expand = set()
1523
items = self.parent_id_basename_to_file_id.iteritems(keys)
1524
next_file_ids = {item[1] for item in items}
1525
next_file_ids = next_file_ids.difference(interesting)
1526
interesting.update(next_file_ids)
1527
for entry in self._getitems(next_file_ids):
1528
if entry.kind == 'directory':
1529
directories_to_expand.add(entry.file_id)
1530
children_of_parent_id.setdefault(entry.parent_id, set()
1531
).add(entry.file_id)
1532
return interesting, children_of_parent_id
1534
def filter(self, specific_fileids):
1535
"""Get an inventory view filtered against a set of file-ids.
1537
Children of directories and parents are included.
1539
The result may or may not reference the underlying inventory
1540
so it should be treated as immutable.
1543
parent_to_children) = self._expand_fileids_to_parents_and_children(
1545
# There is some overlap here, but we assume that all interesting items
1546
# are in the _fileid_to_entry_cache because we had to read them to
1547
# determine if they were a dir we wanted to recurse, or just a file
1548
# This should give us all the entries we'll want to add, so start
1550
other = Inventory(self.root_id)
1551
other.root.revision = self.root.revision
1552
other.revision_id = self.revision_id
1553
if not interesting or not parent_to_children:
1554
# empty filter, or filtering entrys that don't exist
1555
# (if even 1 existed, then we would have populated
1556
# parent_to_children with at least the tree root.)
1558
cache = self._fileid_to_entry_cache
1559
remaining_children = collections.deque(parent_to_children[self.root_id])
1560
while remaining_children:
1561
file_id = remaining_children.popleft()
1563
if ie.kind == 'directory':
1564
ie = ie.copy() # We create a copy to depopulate the .children attribute
1565
# TODO: depending on the uses of 'other' we should probably alwyas
1566
# '.copy()' to prevent someone from mutating other and
1567
# invaliding our internal cache
1569
if file_id in parent_to_children:
1570
remaining_children.extend(parent_to_children[file_id])
1574
def _bytes_to_utf8name_key(data):
1575
"""Get the file_id, revision_id key out of data."""
1576
# We don't normally care about name, except for times when we want
1577
# to filter out empty names because of non rich-root...
1578
sections = data.split(b'\n')
1579
kind, file_id = sections[0].split(b': ')
1580
return (sections[2], bytesintern(file_id), bytesintern(sections[3]))
1582
def _bytes_to_entry(self, bytes):
1583
"""Deserialise a serialised entry."""
1584
sections = bytes.split(b'\n')
1585
if sections[0].startswith(b"file: "):
1586
result = InventoryFile(sections[0][6:],
1587
sections[2].decode('utf8'),
1589
result.text_sha1 = sections[4]
1590
result.text_size = int(sections[5])
1591
result.executable = sections[6] == b"Y"
1592
elif sections[0].startswith(b"dir: "):
1593
result = CHKInventoryDirectory(sections[0][5:],
1594
sections[2].decode('utf8'),
1596
elif sections[0].startswith(b"symlink: "):
1597
result = InventoryLink(sections[0][9:],
1598
sections[2].decode('utf8'),
1600
result.symlink_target = sections[4].decode('utf8')
1601
elif sections[0].startswith(b"tree: "):
1602
result = TreeReference(sections[0][6:],
1603
sections[2].decode('utf8'),
1605
result.reference_revision = sections[4]
1607
raise ValueError("Not a serialised entry %r" % bytes)
1608
result.file_id = bytesintern(result.file_id)
1609
result.revision = bytesintern(sections[3])
1610
if result.parent_id == b'':
1611
result.parent_id = None
1612
self._fileid_to_entry_cache[result.file_id] = result
1615
def create_by_apply_delta(self, inventory_delta, new_revision_id,
1616
propagate_caches=False):
1617
"""Create a new CHKInventory by applying inventory_delta to this one.
1619
See the inventory developers documentation for the theory behind
1622
:param inventory_delta: The inventory delta to apply. See
1623
Inventory.apply_delta for details.
1624
:param new_revision_id: The revision id of the resulting CHKInventory.
1625
:param propagate_caches: If True, the caches for this inventory are
1626
copied to and updated for the result.
1627
:return: The new CHKInventory.
1629
split = osutils.split
1630
result = CHKInventory(self._search_key_name)
1631
if propagate_caches:
1632
# Just propagate the path-to-fileid cache for now
1633
result._path_to_fileid_cache = self._path_to_fileid_cache.copy()
1634
search_key_func = chk_map.search_key_registry.get(self._search_key_name)
1635
self.id_to_entry._ensure_root()
1636
maximum_size = self.id_to_entry._root_node.maximum_size
1637
result.revision_id = new_revision_id
1638
result.id_to_entry = chk_map.CHKMap(
1639
self.id_to_entry._store,
1640
self.id_to_entry.key(),
1641
search_key_func=search_key_func)
1642
result.id_to_entry._ensure_root()
1643
result.id_to_entry._root_node.set_maximum_size(maximum_size)
1644
# Change to apply to the parent_id_basename delta. The dict maps
1645
# (parent_id, basename) -> (old_key, new_value). We use a dict because
1646
# when a path has its id replaced (e.g. the root is changed, or someone
1647
# does bzr mv a b, bzr mv c a, we should output a single change to this
1648
# map rather than two.
1649
parent_id_basename_delta = {}
1650
if self.parent_id_basename_to_file_id is not None:
1651
result.parent_id_basename_to_file_id = chk_map.CHKMap(
1652
self.parent_id_basename_to_file_id._store,
1653
self.parent_id_basename_to_file_id.key(),
1654
search_key_func=search_key_func)
1655
result.parent_id_basename_to_file_id._ensure_root()
1656
self.parent_id_basename_to_file_id._ensure_root()
1657
result_p_id_root = result.parent_id_basename_to_file_id._root_node
1658
p_id_root = self.parent_id_basename_to_file_id._root_node
1659
result_p_id_root.set_maximum_size(p_id_root.maximum_size)
1660
result_p_id_root._key_width = p_id_root._key_width
1662
result.parent_id_basename_to_file_id = None
1663
result.root_id = self.root_id
1664
id_to_entry_delta = []
1665
# inventory_delta is only traversed once, so we just update the
1667
# Check for repeated file ids
1668
inventory_delta = _check_delta_unique_ids(inventory_delta)
1669
# Repeated old paths
1670
inventory_delta = _check_delta_unique_old_paths(inventory_delta)
1671
# Check for repeated new paths
1672
inventory_delta = _check_delta_unique_new_paths(inventory_delta)
1673
# Check for entries that don't match the fileid
1674
inventory_delta = _check_delta_ids_match_entry(inventory_delta)
1675
# Check for nonsense fileids
1676
inventory_delta = _check_delta_ids_are_valid(inventory_delta)
1677
# Check for new_path <-> entry consistency
1678
inventory_delta = _check_delta_new_path_entry_both_or_None(
1680
# All changed entries need to have their parents be directories and be
1681
# at the right path. This set contains (path, id) tuples.
1683
# When we delete an item, all the children of it must be either deleted
1684
# or altered in their own right. As we batch process the change via
1685
# CHKMap.apply_delta, we build a set of things to use to validate the
1689
for old_path, new_path, file_id, entry in inventory_delta:
1692
result.root_id = file_id
1693
if new_path is None:
1698
if propagate_caches:
1700
del result._path_to_fileid_cache[old_path]
1703
deletes.add(file_id)
1705
new_key = StaticTuple(file_id,)
1706
new_value = result._entry_to_bytes(entry)
1707
# Update caches. It's worth doing this whether
1708
# we're propagating the old caches or not.
1709
result._path_to_fileid_cache[new_path] = file_id
1710
parents.add((split(new_path)[0], entry.parent_id))
1711
if old_path is None:
1714
old_key = StaticTuple(file_id,)
1715
if self.id2path(file_id) != old_path:
1716
raise errors.InconsistentDelta(old_path, file_id,
1717
"Entry was at wrong other path %r." %
1718
self.id2path(file_id))
1719
altered.add(file_id)
1720
id_to_entry_delta.append(StaticTuple(old_key, new_key, new_value))
1721
if result.parent_id_basename_to_file_id is not None:
1722
# parent_id, basename changes
1723
if old_path is None:
1726
old_entry = self.get_entry(file_id)
1727
old_key = self._parent_id_basename_key(old_entry)
1728
if new_path is None:
1732
new_key = self._parent_id_basename_key(entry)
1734
# If the two keys are the same, the value will be unchanged
1735
# as its always the file id for this entry.
1736
if old_key != new_key:
1737
# Transform a change into explicit delete/add preserving
1738
# a possible match on the key from a different file id.
1739
if old_key is not None:
1740
parent_id_basename_delta.setdefault(
1741
old_key, [None, None])[0] = old_key
1742
if new_key is not None:
1743
parent_id_basename_delta.setdefault(
1744
new_key, [None, None])[1] = new_value
1745
# validate that deletes are complete.
1746
for file_id in deletes:
1747
entry = self.get_entry(file_id)
1748
if entry.kind != 'directory':
1750
# This loop could potentially be better by using the id_basename
1751
# map to just get the child file ids.
1752
for child in viewvalues(entry.children):
1753
if child.file_id not in altered:
1754
raise errors.InconsistentDelta(self.id2path(child.file_id),
1755
child.file_id, "Child not deleted or reparented when "
1757
result.id_to_entry.apply_delta(id_to_entry_delta)
1758
if parent_id_basename_delta:
1759
# Transform the parent_id_basename delta data into a linear delta
1760
# with only one record for a given key. Optimally this would allow
1761
# re-keying, but its simpler to just output that as a delete+add
1762
# to spend less time calculating the delta.
1764
for key, (old_key, value) in viewitems(parent_id_basename_delta):
1765
if value is not None:
1766
delta_list.append((old_key, key, value))
1768
delta_list.append((old_key, None, None))
1769
result.parent_id_basename_to_file_id.apply_delta(delta_list)
1770
parents.discard(('', None))
1771
for parent_path, parent in parents:
1773
if result.get_entry(parent).kind != 'directory':
1774
raise errors.InconsistentDelta(result.id2path(parent), parent,
1775
'Not a directory, but given children')
1776
except errors.NoSuchId:
1777
raise errors.InconsistentDelta("<unknown>", parent,
1778
"Parent is not present in resulting inventory.")
1779
if result.path2id(parent_path) != parent:
1780
raise errors.InconsistentDelta(parent_path, parent,
1781
"Parent has wrong path %r." % result.path2id(parent_path))
1785
def deserialise(klass, chk_store, bytes, expected_revision_id):
1786
"""Deserialise a CHKInventory.
1788
:param chk_store: A CHK capable VersionedFiles instance.
1789
:param bytes: The serialised bytes.
1790
:param expected_revision_id: The revision ID we think this inventory is
1792
:return: A CHKInventory
1794
lines = bytes.split(b'\n')
1795
if lines[-1] != b'':
1796
raise AssertionError('bytes to deserialize must end with an eol')
1798
if lines[0] != b'chkinventory:':
1799
raise ValueError("not a serialised CHKInventory: %r" % bytes)
1801
allowed_keys = frozenset((b'root_id', b'revision_id',
1802
b'parent_id_basename_to_file_id',
1803
b'search_key_name', b'id_to_entry'))
1804
for line in lines[1:]:
1805
key, value = line.split(b': ', 1)
1806
if key not in allowed_keys:
1807
raise errors.BzrError('Unknown key in inventory: %r\n%r'
1810
raise errors.BzrError('Duplicate key in inventory: %r\n%r'
1813
revision_id = bytesintern(info[b'revision_id'])
1814
root_id = bytesintern(info[b'root_id'])
1815
search_key_name = bytesintern(info.get(b'search_key_name', b'plain'))
1816
parent_id_basename_to_file_id = bytesintern(info.get(
1817
b'parent_id_basename_to_file_id', None))
1818
if not parent_id_basename_to_file_id.startswith(b'sha1:'):
1819
raise ValueError('parent_id_basename_to_file_id should be a sha1'
1820
' key not %r' % (parent_id_basename_to_file_id,))
1821
id_to_entry = info[b'id_to_entry']
1822
if not id_to_entry.startswith(b'sha1:'):
1823
raise ValueError('id_to_entry should be a sha1'
1824
' key not %r' % (id_to_entry,))
1826
result = CHKInventory(search_key_name)
1827
result.revision_id = revision_id
1828
result.root_id = root_id
1829
search_key_func = chk_map.search_key_registry.get(
1830
result._search_key_name)
1831
if parent_id_basename_to_file_id is not None:
1832
result.parent_id_basename_to_file_id = chk_map.CHKMap(
1833
chk_store, StaticTuple(parent_id_basename_to_file_id,),
1834
search_key_func=search_key_func)
1836
result.parent_id_basename_to_file_id = None
1838
result.id_to_entry = chk_map.CHKMap(chk_store,
1839
StaticTuple(id_to_entry,),
1840
search_key_func=search_key_func)
1841
if (result.revision_id,) != expected_revision_id:
1842
raise ValueError("Mismatched revision id and expected: %r, %r" %
1843
(result.revision_id, expected_revision_id))
1847
def from_inventory(klass, chk_store, inventory, maximum_size=0, search_key_name=b'plain'):
1848
"""Create a CHKInventory from an existing inventory.
1850
The content of inventory is copied into the chk_store, and a
1851
CHKInventory referencing that is returned.
1853
:param chk_store: A CHK capable VersionedFiles instance.
1854
:param inventory: The inventory to copy.
1855
:param maximum_size: The CHKMap node size limit.
1856
:param search_key_name: The identifier for the search key function
1858
result = klass(search_key_name)
1859
result.revision_id = inventory.revision_id
1860
result.root_id = inventory.root.file_id
1862
entry_to_bytes = result._entry_to_bytes
1863
parent_id_basename_key = result._parent_id_basename_key
1864
id_to_entry_dict = {}
1865
parent_id_basename_dict = {}
1866
for path, entry in inventory.iter_entries():
1867
key = StaticTuple(entry.file_id,).intern()
1868
id_to_entry_dict[key] = entry_to_bytes(entry)
1869
p_id_key = parent_id_basename_key(entry)
1870
parent_id_basename_dict[p_id_key] = entry.file_id
1872
result._populate_from_dicts(chk_store, id_to_entry_dict,
1873
parent_id_basename_dict, maximum_size=maximum_size)
1876
def _populate_from_dicts(self, chk_store, id_to_entry_dict,
1877
parent_id_basename_dict, maximum_size):
1878
search_key_func = chk_map.search_key_registry.get(self._search_key_name)
1879
root_key = chk_map.CHKMap.from_dict(chk_store, id_to_entry_dict,
1880
maximum_size=maximum_size, key_width=1,
1881
search_key_func=search_key_func)
1882
self.id_to_entry = chk_map.CHKMap(chk_store, root_key,
1884
root_key = chk_map.CHKMap.from_dict(chk_store,
1885
parent_id_basename_dict,
1886
maximum_size=maximum_size, key_width=2,
1887
search_key_func=search_key_func)
1888
self.parent_id_basename_to_file_id = chk_map.CHKMap(chk_store,
1889
root_key, search_key_func)
1891
def _parent_id_basename_key(self, entry):
1892
"""Create a key for a entry in a parent_id_basename_to_file_id index."""
1893
if entry.parent_id is not None:
1894
parent_id = entry.parent_id
1897
return StaticTuple(parent_id, entry.name.encode('utf8')).intern()
1899
def get_entry(self, file_id):
1900
"""map a single file_id -> InventoryEntry."""
1902
raise errors.NoSuchId(self, file_id)
1903
result = self._fileid_to_entry_cache.get(file_id, None)
1904
if result is not None:
1907
return self._bytes_to_entry(
1908
next(self.id_to_entry.iteritems([StaticTuple(file_id,)]))[1])
1909
except StopIteration:
1910
# really we're passing an inventory, not a tree...
1911
raise errors.NoSuchId(self, file_id)
1913
def _getitems(self, file_ids):
1914
"""Similar to get_entry, but lets you query for multiple.
1916
The returned order is undefined. And currently if an item doesn't
1917
exist, it isn't included in the output.
1921
for file_id in file_ids:
1922
entry = self._fileid_to_entry_cache.get(file_id, None)
1924
remaining.append(file_id)
1926
result.append(entry)
1927
file_keys = [StaticTuple(f,).intern() for f in remaining]
1928
for file_key, value in self.id_to_entry.iteritems(file_keys):
1929
entry = self._bytes_to_entry(value)
1930
result.append(entry)
1931
self._fileid_to_entry_cache[entry.file_id] = entry
1934
def has_id(self, file_id):
1935
# Perhaps have an explicit 'contains' method on CHKMap ?
1936
if self._fileid_to_entry_cache.get(file_id, None) is not None:
1939
self.id_to_entry.iteritems([StaticTuple(file_id,)]))) == 1
1941
def is_root(self, file_id):
1942
return file_id == self.root_id
1944
def _iter_file_id_parents(self, file_id):
1945
"""Yield the parents of file_id up to the root."""
1946
while file_id is not None:
1948
ie = self.get_entry(file_id)
1950
raise errors.NoSuchId(tree=self, file_id=file_id)
1952
file_id = ie.parent_id
1954
def iter_all_ids(self):
1955
"""Iterate over all file-ids."""
1956
for key, _ in self.id_to_entry.iteritems():
1959
def iter_just_entries(self):
1960
"""Iterate over all entries.
1962
Unlike iter_entries(), just the entries are returned (not (path, ie))
1963
and the order of entries is undefined.
1965
XXX: We may not want to merge this into bzr.dev.
1967
for key, entry in self.id_to_entry.iteritems():
1969
ie = self._fileid_to_entry_cache.get(file_id, None)
1971
ie = self._bytes_to_entry(entry)
1972
self._fileid_to_entry_cache[file_id] = ie
1975
def _preload_cache(self):
1976
"""Make sure all file-ids are in _fileid_to_entry_cache"""
1977
if self._fully_cached:
1978
return # No need to do it again
1979
# The optimal sort order is to use iteritems() directly
1980
cache = self._fileid_to_entry_cache
1981
for key, entry in self.id_to_entry.iteritems():
1983
if file_id not in cache:
1984
ie = self._bytes_to_entry(entry)
1988
last_parent_id = last_parent_ie = None
1989
pid_items = self.parent_id_basename_to_file_id.iteritems()
1990
for key, child_file_id in pid_items:
1991
if key == (b'', b''): # This is the root
1992
if child_file_id != self.root_id:
1993
raise ValueError('Data inconsistency detected.'
1994
' We expected data with key ("","") to match'
1995
' the root id, but %s != %s'
1996
% (child_file_id, self.root_id))
1998
parent_id, basename = key
1999
ie = cache[child_file_id]
2000
if parent_id == last_parent_id:
2001
parent_ie = last_parent_ie
2003
parent_ie = cache[parent_id]
2004
if parent_ie.kind != 'directory':
2005
raise ValueError('Data inconsistency detected.'
2006
' An entry in the parent_id_basename_to_file_id map'
2007
' has parent_id {%s} but the kind of that object'
2008
' is %r not "directory"' % (parent_id, parent_ie.kind))
2009
if parent_ie._children is None:
2010
parent_ie._children = {}
2011
basename = basename.decode('utf-8')
2012
if basename in parent_ie._children:
2013
existing_ie = parent_ie._children[basename]
2014
if existing_ie != ie:
2015
raise ValueError('Data inconsistency detected.'
2016
' Two entries with basename %r were found'
2017
' in the parent entry {%s}'
2018
% (basename, parent_id))
2019
if basename != ie.name:
2020
raise ValueError('Data inconsistency detected.'
2021
' In the parent_id_basename_to_file_id map, file_id'
2022
' {%s} is listed as having basename %r, but in the'
2023
' id_to_entry map it is %r'
2024
% (child_file_id, basename, ie.name))
2025
parent_ie._children[basename] = ie
2026
self._fully_cached = True
2028
def iter_changes(self, basis):
2029
"""Generate a Tree.iter_changes change list between this and basis.
2031
:param basis: Another CHKInventory.
2032
:return: An iterator over the changes between self and basis, as per
2033
tree.iter_changes().
2035
# We want: (file_id, (path_in_source, path_in_target),
2036
# changed_content, versioned, parent, name, kind,
2038
for key, basis_value, self_value in \
2039
self.id_to_entry.iter_changes(basis.id_to_entry):
2041
if basis_value is not None:
2042
basis_entry = basis._bytes_to_entry(basis_value)
2043
path_in_source = basis.id2path(file_id)
2044
basis_parent = basis_entry.parent_id
2045
basis_name = basis_entry.name
2046
basis_executable = basis_entry.executable
2048
path_in_source = None
2051
basis_executable = None
2052
if self_value is not None:
2053
self_entry = self._bytes_to_entry(self_value)
2054
path_in_target = self.id2path(file_id)
2055
self_parent = self_entry.parent_id
2056
self_name = self_entry.name
2057
self_executable = self_entry.executable
2059
path_in_target = None
2062
self_executable = None
2063
if basis_value is None:
2065
kind = (None, self_entry.kind)
2066
versioned = (False, True)
2067
elif self_value is None:
2069
kind = (basis_entry.kind, None)
2070
versioned = (True, False)
2072
kind = (basis_entry.kind, self_entry.kind)
2073
versioned = (True, True)
2074
changed_content = False
2075
if kind[0] != kind[1]:
2076
changed_content = True
2077
elif kind[0] == 'file':
2078
if (self_entry.text_size != basis_entry.text_size or
2079
self_entry.text_sha1 != basis_entry.text_sha1):
2080
changed_content = True
2081
elif kind[0] == 'symlink':
2082
if self_entry.symlink_target != basis_entry.symlink_target:
2083
changed_content = True
2084
elif kind[0] == 'tree-reference':
2085
if (self_entry.reference_revision !=
2086
basis_entry.reference_revision):
2087
changed_content = True
2088
parent = (basis_parent, self_parent)
2089
name = (basis_name, self_name)
2090
executable = (basis_executable, self_executable)
2091
if (not changed_content
2092
and parent[0] == parent[1]
2093
and name[0] == name[1]
2094
and executable[0] == executable[1]):
2095
# Could happen when only the revision changed for a directory
2098
yield (file_id, (path_in_source, path_in_target), changed_content,
2099
versioned, parent, name, kind, executable)
2102
"""Return the number of entries in the inventory."""
2103
return len(self.id_to_entry)
2105
def _make_delta(self, old):
2106
"""Make an inventory delta from two inventories."""
2107
if not isinstance(old, CHKInventory):
2108
return CommonInventory._make_delta(self, old)
2110
for key, old_value, self_value in \
2111
self.id_to_entry.iter_changes(old.id_to_entry):
2113
if old_value is not None:
2114
old_path = old.id2path(file_id)
2117
if self_value is not None:
2118
entry = self._bytes_to_entry(self_value)
2119
self._fileid_to_entry_cache[file_id] = entry
2120
new_path = self.id2path(file_id)
2124
delta.append((old_path, new_path, file_id, entry))
2127
def path2id(self, relpath):
2128
"""See CommonInventory.path2id()."""
2129
# TODO: perhaps support negative hits?
2130
if isinstance(relpath, (str, text_type)):
2131
names = osutils.splitpath(relpath)
2136
relpath = osutils.pathjoin(*relpath)
2137
result = self._path_to_fileid_cache.get(relpath, None)
2138
if result is not None:
2140
current_id = self.root_id
2141
if current_id is None:
2143
parent_id_index = self.parent_id_basename_to_file_id
2145
for basename in names:
2146
if cur_path is None:
2149
cur_path = cur_path + '/' + basename
2150
basename_utf8 = basename.encode('utf8')
2151
file_id = self._path_to_fileid_cache.get(cur_path, None)
2153
key_filter = [StaticTuple(current_id, basename_utf8)]
2154
items = parent_id_index.iteritems(key_filter)
2155
for (parent_id, name_utf8), file_id in items:
2156
if parent_id != current_id or name_utf8 != basename_utf8:
2157
raise errors.BzrError("corrupt inventory lookup! "
2158
"%r %r %r %r" % (parent_id, current_id, name_utf8,
2163
self._path_to_fileid_cache[cur_path] = file_id
2164
current_id = file_id
2168
"""Serialise the inventory to lines."""
2169
lines = [b"chkinventory:\n"]
2170
if self._search_key_name != b'plain':
2171
# custom ordering grouping things that don't change together
2172
lines.append(b'search_key_name: %s\n' % (
2173
self._search_key_name))
2174
lines.append(b"root_id: %s\n" % self.root_id)
2175
lines.append(b'parent_id_basename_to_file_id: %s\n' %
2176
(self.parent_id_basename_to_file_id.key()[0],))
2177
lines.append(b"revision_id: %s\n" % self.revision_id)
2178
lines.append(b"id_to_entry: %s\n" % (self.id_to_entry.key()[0],))
2180
lines.append(b"revision_id: %s\n" % self.revision_id)
2181
lines.append(b"root_id: %s\n" % self.root_id)
2182
if self.parent_id_basename_to_file_id is not None:
2183
lines.append(b'parent_id_basename_to_file_id: %s\n' %
2184
(self.parent_id_basename_to_file_id.key()[0],))
2185
lines.append(b"id_to_entry: %s\n" % (self.id_to_entry.key()[0],))
2190
"""Get the root entry."""
2191
return self.get_entry(self.root_id)
2194
class CHKInventoryDirectory(InventoryDirectory):
2195
"""A directory in an inventory."""
2197
__slots__ = ['_children', '_chk_inventory']
2199
def __init__(self, file_id, name, parent_id, chk_inventory):
2200
# Don't call InventoryDirectory.__init__ - it isn't right for this
2202
InventoryEntry.__init__(self, file_id, name, parent_id)
2203
self._children = None
2204
self._chk_inventory = chk_inventory
2208
"""Access the list of children of this directory.
2210
With a parent_id_basename_to_file_id index, loads all the children,
2211
without loads the entire index. Without is bad. A more sophisticated
2212
proxy object might be nice, to allow partial loading of children as
2213
well when specific names are accessed. (So path traversal can be
2214
written in the obvious way but not examine siblings.).
2216
if self._children is not None:
2217
return self._children
2218
# No longer supported
2219
if self._chk_inventory.parent_id_basename_to_file_id is None:
2220
raise AssertionError("Inventories without"
2221
" parent_id_basename_to_file_id are no longer supported")
2223
# XXX: Todo - use proxy objects for the children rather than loading
2224
# all when the attribute is referenced.
2225
parent_id_index = self._chk_inventory.parent_id_basename_to_file_id
2227
for (parent_id, name_utf8), file_id in parent_id_index.iteritems(
2228
key_filter=[StaticTuple(self.file_id,)]):
2229
child_keys.add(StaticTuple(file_id,))
2231
for file_id_key in child_keys:
2232
entry = self._chk_inventory._fileid_to_entry_cache.get(
2233
file_id_key[0], None)
2234
if entry is not None:
2235
result[entry.name] = entry
2236
cached.add(file_id_key)
2237
child_keys.difference_update(cached)
2238
# populate; todo: do by name
2239
id_to_entry = self._chk_inventory.id_to_entry
2240
for file_id_key, bytes in id_to_entry.iteritems(child_keys):
2241
entry = self._chk_inventory._bytes_to_entry(bytes)
2242
result[entry.name] = entry
2243
self._chk_inventory._fileid_to_entry_cache[file_id_key[0]] = entry
2244
self._children = result
2248
'directory': InventoryDirectory,
2249
'file': InventoryFile,
2250
'symlink': InventoryLink,
2251
'tree-reference': TreeReference
2254
def make_entry(kind, name, parent_id, file_id=None):
2255
"""Create an inventory entry.
2257
:param kind: the type of inventory entry to create.
2258
:param name: the basename of the entry.
2259
:param parent_id: the parent_id of the entry.
2260
:param file_id: the file_id to use. if None, one will be created.
2263
file_id = generate_ids.gen_file_id(name)
2264
name = ensure_normalized_name(name)
2266
factory = entry_factory[kind]
2268
raise errors.BadFileKindError(name, kind)
2269
return factory(file_id, name, parent_id)
2272
def ensure_normalized_name(name):
2275
:raises InvalidNormalization: When name is not normalized, and cannot be
2276
accessed on this platform by the normalized path.
2277
:return: The NFC normalised version of name.
2279
#------- This has been copied to breezy.dirstate.DirState.add, please
2280
# keep them synchronised.
2281
# we dont import normalized_filename directly because we want to be
2282
# able to change the implementation at runtime for tests.
2283
norm_name, can_access = osutils.normalized_filename(name)
2284
if norm_name != name:
2288
# TODO: jam 20060701 This would probably be more useful
2289
# if the error was raised with the full path
2290
raise errors.InvalidNormalization(name)
2294
_NAME_RE = lazy_regex.lazy_compile(r'^[^/\\]+$')
2296
def is_valid_name(name):
2297
return bool(_NAME_RE.match(name))
2300
def _check_delta_unique_ids(delta):
2301
"""Decorate a delta and check that the file ids in it are unique.
2303
:return: A generator over delta.
2307
length = len(ids) + 1
2309
if len(ids) != length:
2310
raise errors.InconsistentDelta(item[0] or item[1], item[2],
2315
def _check_delta_unique_new_paths(delta):
2316
"""Decorate a delta and check that the new paths in it are unique.
2318
:return: A generator over delta.
2322
length = len(paths) + 1
2324
if path is not None:
2326
if len(paths) != length:
2327
raise errors.InconsistentDelta(path, item[2], "repeated path")
2331
def _check_delta_unique_old_paths(delta):
2332
"""Decorate a delta and check that the old paths in it are unique.
2334
:return: A generator over delta.
2338
length = len(paths) + 1
2340
if path is not None:
2342
if len(paths) != length:
2343
raise errors.InconsistentDelta(path, item[2], "repeated path")
2347
def _check_delta_ids_are_valid(delta):
2348
"""Decorate a delta and check that the ids in it are valid.
2350
:return: A generator over delta.
2355
raise errors.InconsistentDelta(item[0] or item[1], item[2],
2356
"entry with file_id None %r" % entry)
2357
if not isinstance(item[2], bytes):
2358
raise errors.InconsistentDelta(item[0] or item[1], item[2],
2359
"entry with non bytes file_id %r" % entry)
2363
def _check_delta_ids_match_entry(delta):
2364
"""Decorate a delta and check that the ids in it match the entry.file_id.
2366
:return: A generator over delta.
2370
if entry is not None:
2371
if entry.file_id != item[2]:
2372
raise errors.InconsistentDelta(item[0] or item[1], item[2],
2373
"mismatched id with %r" % entry)
2377
def _check_delta_new_path_entry_both_or_None(delta):
2378
"""Decorate a delta and check that the new_path and entry are paired.
2380
:return: A generator over delta.
2385
if new_path is None and entry is not None:
2386
raise errors.InconsistentDelta(item[0], item[1],
2387
"Entry with no new_path")
2388
if new_path is not None and entry is None:
2389
raise errors.InconsistentDelta(new_path, item[1],
2390
"new_path with no entry")
2394
def mutable_inventory_from_tree(tree):
2395
"""Create a new inventory that has the same contents as a specified tree.
2397
:param tree: Revision tree to create inventory from
2399
entries = tree.iter_entries_by_dir()
2400
inv = Inventory(None, tree.get_revision_id())
2401
for path, inv_entry in entries:
2402
inv.add(inv_entry.copy())
added
removed
breezy/bzr/inventory.py
