1
# Copyright (C) 2005, 2006, 2007 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
# This should really be an id randomly assigned when the tree is
27
# created, but it's not for now.
30
from bzrlib.lazy_import import lazy_import
31
lazy_import(globals(), """
48
from bzrlib.errors import (
52
from bzrlib.symbol_versioning import deprecated_in, deprecated_method
53
from bzrlib.trace import mutter
56
class InventoryEntry(object):
57
"""Description of a versioned file.
59
An InventoryEntry has the following fields, which are also
60
present in the XML inventory-entry element:
65
(within the parent directory)
68
file_id of the parent directory, or ROOT_ID
71
the revision_id in which this variation of this file was
75
Indicates that this file should be executable on systems
79
sha-1 of the text of the file
82
size in bytes of the text of the file
84
(reading a version 4 tree created a text_id field.)
89
>>> i.add(InventoryDirectory('123', 'src', ROOT_ID))
90
InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None)
91
>>> i.add(InventoryFile('2323', 'hello.c', parent_id='123'))
92
InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None, revision=None)
93
>>> shouldbe = {0: '', 1: 'src', 2: 'src/hello.c'}
94
>>> for ix, j in enumerate(i.iter_entries()):
95
... print (j[0] == shouldbe[ix], j[1])
97
(True, InventoryDirectory('TREE_ROOT', u'', parent_id=None, revision=None))
98
(True, InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None))
99
(True, InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None, revision=None))
100
>>> i.add(InventoryFile('2324', 'bye.c', '123'))
101
InventoryFile('2324', 'bye.c', parent_id='123', sha1=None, len=None, revision=None)
102
>>> i.add(InventoryDirectory('2325', 'wibble', '123'))
103
InventoryDirectory('2325', 'wibble', parent_id='123', revision=None)
104
>>> i.path2id('src/wibble')
108
>>> i.add(InventoryFile('2326', 'wibble.c', '2325'))
109
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None, revision=None)
111
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None, revision=None)
112
>>> for path, entry in i.iter_entries():
121
>>> i.id2path('2326')
122
'src/wibble/wibble.c'
125
# Constants returned by describe_change()
127
# TODO: These should probably move to some kind of FileChangeDescription
128
# class; that's like what's inside a TreeDelta but we want to be able to
129
# generate them just for one file at a time.
131
MODIFIED_AND_RENAMED = 'modified and renamed'
135
def detect_changes(self, old_entry):
136
"""Return a (text_modified, meta_modified) from this to old_entry.
138
_read_tree_state must have been called on self and old_entry prior to
139
calling detect_changes.
143
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
144
output_to, reverse=False):
145
"""Perform a diff between two entries of the same kind."""
147
def parent_candidates(self, previous_inventories):
148
"""Find possible per-file graph parents.
150
This is currently defined by:
151
- Select the last changed revision in the parent inventory.
152
- Do deal with a short lived bug in bzr 0.8's development two entries
153
that have the same last changed but different 'x' bit settings are
156
# revision:ie mapping for each ie found in previous_inventories.
158
# identify candidate head revision ids.
159
for inv in previous_inventories:
160
if self.file_id in inv:
161
ie = inv[self.file_id]
162
if ie.revision in candidates:
163
# same revision value in two different inventories:
164
# correct possible inconsistencies:
165
# * there was a bug in revision updates with 'x' bit
168
if candidates[ie.revision].executable != ie.executable:
169
candidates[ie.revision].executable = False
170
ie.executable = False
171
except AttributeError:
174
# add this revision as a candidate.
175
candidates[ie.revision] = ie
178
@deprecated_method(deprecated_in((1, 6, 0)))
179
def get_tar_item(self, root, dp, now, tree):
180
"""Get a tarfile item and a file stream for its content."""
181
item = tarfile.TarInfo(osutils.pathjoin(root, dp).encode('utf8'))
182
# TODO: would be cool to actually set it to the timestamp of the
183
# revision it was last changed
185
fileobj = self._put_in_tar(item, tree)
189
"""Return true if the object this entry represents has textual data.
191
Note that textual data includes binary content.
193
Also note that all entries get weave files created for them.
194
This attribute is primarily used when upgrading from old trees that
195
did not have the weave index for all inventory entries.
199
def __init__(self, file_id, name, parent_id, text_id=None):
200
"""Create an InventoryEntry
202
The filename must be a single component, relative to the
203
parent directory; it cannot be a whole path or relative name.
205
>>> e = InventoryFile('123', 'hello.c', ROOT_ID)
210
>>> e = InventoryFile('123', 'src/hello.c', ROOT_ID)
211
Traceback (most recent call last):
212
InvalidEntryName: Invalid entry name: src/hello.c
214
if '/' in name or '\\' in name:
215
raise errors.InvalidEntryName(name=name)
216
self.executable = False
218
self.text_sha1 = None
219
self.text_size = None
220
self.file_id = file_id
222
self.text_id = text_id
223
self.parent_id = parent_id
224
self.symlink_target = None
225
self.reference_revision = None
227
def kind_character(self):
228
"""Return a short kind indicator useful for appending to names."""
229
raise BzrError('unknown kind %r' % self.kind)
231
known_kinds = ('file', 'directory', 'symlink')
233
def _put_in_tar(self, item, tree):
234
"""populate item for stashing in a tar, and return the content stream.
236
If no content is available, return None.
238
raise BzrError("don't know how to export {%s} of kind %r" %
239
(self.file_id, self.kind))
241
@deprecated_method(deprecated_in((1, 6, 0)))
242
def put_on_disk(self, dest, dp, tree):
243
"""Create a representation of self on disk in the prefix dest.
245
This is a template method - implement _put_on_disk in subclasses.
247
fullpath = osutils.pathjoin(dest, dp)
248
self._put_on_disk(fullpath, tree)
249
# mutter(" export {%s} kind %s to %s", self.file_id,
250
# self.kind, fullpath)
252
def _put_on_disk(self, fullpath, tree):
253
"""Put this entry onto disk at fullpath, from tree tree."""
254
raise BzrError("don't know how to export {%s} of kind %r" % (self.file_id, self.kind))
256
def sorted_children(self):
257
return sorted(self.children.items())
260
def versionable_kind(kind):
261
return (kind in ('file', 'directory', 'symlink', 'tree-reference'))
263
def check(self, checker, rev_id, inv, tree):
264
"""Check this inventory entry is intact.
266
This is a template method, override _check for kind specific
269
:param checker: Check object providing context for the checks;
270
can be used to find out what parts of the repository have already
272
:param rev_id: Revision id from which this InventoryEntry was loaded.
273
Not necessarily the last-changed revision for this file.
274
:param inv: Inventory from which the entry was loaded.
275
:param tree: RevisionTree for this entry.
277
if self.parent_id is not None:
278
if not inv.has_id(self.parent_id):
279
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
280
% (self.parent_id, rev_id))
281
self._check(checker, rev_id, tree)
283
def _check(self, checker, rev_id, tree):
284
"""Check this inventory entry for kind specific errors."""
285
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
289
"""Clone this inventory entry."""
290
raise NotImplementedError
293
def describe_change(old_entry, new_entry):
294
"""Describe the change between old_entry and this.
296
This smells of being an InterInventoryEntry situation, but as its
297
the first one, we're making it a static method for now.
299
An entry with a different parent, or different name is considered
300
to be renamed. Reparenting is an internal detail.
301
Note that renaming the parent does not trigger a rename for the
304
# TODO: Perhaps return an object rather than just a string
305
if old_entry is new_entry:
306
# also the case of both being None
308
elif old_entry is None:
310
elif new_entry is None:
312
if old_entry.kind != new_entry.kind:
314
text_modified, meta_modified = new_entry.detect_changes(old_entry)
315
if text_modified or meta_modified:
319
# TODO 20060511 (mbp, rbc) factor out 'detect_rename' here.
320
if old_entry.parent_id != new_entry.parent_id:
322
elif old_entry.name != new_entry.name:
326
if renamed and not modified:
327
return InventoryEntry.RENAMED
328
if modified and not renamed:
330
if modified and renamed:
331
return InventoryEntry.MODIFIED_AND_RENAMED
335
return ("%s(%r, %r, parent_id=%r, revision=%r)"
336
% (self.__class__.__name__,
342
def __eq__(self, other):
344
# For the case when objects are cached
346
if not isinstance(other, InventoryEntry):
347
return NotImplemented
349
return ((self.file_id == other.file_id)
350
and (self.name == other.name)
351
and (other.symlink_target == self.symlink_target)
352
and (self.text_sha1 == other.text_sha1)
353
and (self.text_size == other.text_size)
354
and (self.text_id == other.text_id)
355
and (self.parent_id == other.parent_id)
356
and (self.kind == other.kind)
357
and (self.revision == other.revision)
358
and (self.executable == other.executable)
359
and (self.reference_revision == other.reference_revision)
362
def __ne__(self, other):
363
return not (self == other)
366
raise ValueError('not hashable')
368
def _unchanged(self, previous_ie):
369
"""Has this entry changed relative to previous_ie.
371
This method should be overridden in child classes.
374
# different inv parent
375
if previous_ie.parent_id != self.parent_id:
378
elif previous_ie.name != self.name:
380
elif previous_ie.kind != self.kind:
384
def _read_tree_state(self, path, work_tree):
385
"""Populate fields in the inventory entry from the given tree.
387
Note that this should be modified to be a noop on virtual trees
388
as all entries created there are prepopulated.
390
# TODO: Rather than running this manually, we should check the
391
# working sha1 and other expensive properties when they're
392
# first requested, or preload them if they're already known
393
pass # nothing to do by default
395
def _forget_tree_state(self):
399
class RootEntry(InventoryEntry):
401
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
402
'text_id', 'parent_id', 'children', 'executable',
403
'revision', 'symlink_target', 'reference_revision']
405
def _check(self, checker, rev_id, tree):
406
"""See InventoryEntry._check"""
408
def __init__(self, file_id):
409
self.file_id = file_id
411
self.kind = 'directory'
412
self.parent_id = None
415
symbol_versioning.warn('RootEntry is deprecated as of bzr 0.10.'
416
' Please use InventoryDirectory instead.',
417
DeprecationWarning, stacklevel=2)
419
def __eq__(self, other):
420
if not isinstance(other, RootEntry):
421
return NotImplemented
423
return (self.file_id == other.file_id) \
424
and (self.children == other.children)
427
class InventoryDirectory(InventoryEntry):
428
"""A directory in an inventory."""
430
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
431
'text_id', 'parent_id', 'children', 'executable',
432
'revision', 'symlink_target', 'reference_revision']
434
def _check(self, checker, rev_id, tree):
435
"""See InventoryEntry._check"""
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}'
438
% (self.file_id, rev_id))
441
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
442
other.revision = self.revision
443
# note that children are *not* copied; they're pulled across when
447
def __init__(self, file_id, name, parent_id):
448
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
450
self.kind = 'directory'
452
def kind_character(self):
453
"""See InventoryEntry.kind_character."""
456
def _put_in_tar(self, item, tree):
457
"""See InventoryEntry._put_in_tar."""
458
item.type = tarfile.DIRTYPE
465
def _put_on_disk(self, fullpath, tree):
466
"""See InventoryEntry._put_on_disk."""
470
class InventoryFile(InventoryEntry):
471
"""A file in an inventory."""
473
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
474
'text_id', 'parent_id', 'children', 'executable',
475
'revision', 'symlink_target', 'reference_revision']
477
def _check(self, checker, tree_revision_id, tree):
478
"""See InventoryEntry._check"""
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
500
other = InventoryFile(self.file_id, self.name, self.parent_id)
501
other.executable = self.executable
502
other.text_id = self.text_id
503
other.text_sha1 = self.text_sha1
504
other.text_size = self.text_size
505
other.revision = self.revision
508
def detect_changes(self, old_entry):
509
"""See InventoryEntry.detect_changes."""
510
text_modified = (self.text_sha1 != old_entry.text_sha1)
511
meta_modified = (self.executable != old_entry.executable)
512
return text_modified, meta_modified
514
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
515
output_to, reverse=False):
516
"""See InventoryEntry._diff."""
517
from bzrlib.diff import DiffText
518
from_file_id = self.file_id
520
to_file_id = to_entry.file_id
524
to_file_id, from_file_id = from_file_id, to_file_id
525
tree, to_tree = to_tree, tree
526
from_label, to_label = to_label, from_label
527
differ = DiffText(tree, to_tree, output_to, 'utf-8', '', '',
529
return differ.diff_text(from_file_id, to_file_id, from_label, to_label)
532
"""See InventoryEntry.has_text."""
535
def __init__(self, file_id, name, parent_id):
536
super(InventoryFile, self).__init__(file_id, name, parent_id)
539
def kind_character(self):
540
"""See InventoryEntry.kind_character."""
543
def _put_in_tar(self, item, tree):
544
"""See InventoryEntry._put_in_tar."""
545
item.type = tarfile.REGTYPE
546
fileobj = tree.get_file(self.file_id)
547
item.size = self.text_size
548
if tree.is_executable(self.file_id):
554
def _put_on_disk(self, fullpath, tree):
555
"""See InventoryEntry._put_on_disk."""
556
osutils.pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
557
if tree.is_executable(self.file_id):
558
os.chmod(fullpath, 0755)
560
def _read_tree_state(self, path, work_tree):
561
"""See InventoryEntry._read_tree_state."""
562
self.text_sha1 = work_tree.get_file_sha1(self.file_id, path=path)
563
# FIXME: 20050930 probe for the text size when getting sha1
564
# in _read_tree_state
565
self.executable = work_tree.is_executable(self.file_id, path=path)
568
return ("%s(%r, %r, parent_id=%r, sha1=%r, len=%s, revision=%s)"
569
% (self.__class__.__name__,
577
def _forget_tree_state(self):
578
self.text_sha1 = None
580
def _unchanged(self, previous_ie):
581
"""See InventoryEntry._unchanged."""
582
compatible = super(InventoryFile, self)._unchanged(previous_ie)
583
if self.text_sha1 != previous_ie.text_sha1:
586
# FIXME: 20050930 probe for the text size when getting sha1
587
# in _read_tree_state
588
self.text_size = previous_ie.text_size
589
if self.executable != previous_ie.executable:
594
class InventoryLink(InventoryEntry):
595
"""A file in an inventory."""
597
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
598
'text_id', 'parent_id', 'children', 'executable',
599
'revision', 'symlink_target', 'reference_revision']
601
def _check(self, checker, rev_id, tree):
602
"""See InventoryEntry._check"""
603
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
604
raise BzrCheckError('symlink {%s} has text in revision {%s}'
605
% (self.file_id, rev_id))
606
if self.symlink_target is None:
607
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
608
% (self.file_id, rev_id))
611
other = InventoryLink(self.file_id, self.name, self.parent_id)
612
other.symlink_target = self.symlink_target
613
other.revision = self.revision
616
def detect_changes(self, old_entry):
617
"""See InventoryEntry.detect_changes."""
618
# FIXME: which _modified field should we use ? RBC 20051003
619
text_modified = (self.symlink_target != old_entry.symlink_target)
621
mutter(" symlink target changed")
622
meta_modified = False
623
return text_modified, meta_modified
625
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
626
output_to, reverse=False):
627
"""See InventoryEntry._diff."""
628
from bzrlib.diff import DiffSymlink
629
old_target = self.symlink_target
630
if to_entry is not None:
631
new_target = to_entry.symlink_target
640
new_target, old_target = old_target, new_target
641
differ = DiffSymlink(old_tree, new_tree, output_to)
642
return differ.diff_symlink(old_target, new_target)
644
def __init__(self, file_id, name, parent_id):
645
super(InventoryLink, self).__init__(file_id, name, parent_id)
646
self.kind = 'symlink'
648
def kind_character(self):
649
"""See InventoryEntry.kind_character."""
652
def _put_in_tar(self, item, tree):
653
"""See InventoryEntry._put_in_tar."""
654
item.type = tarfile.SYMTYPE
658
item.linkname = self.symlink_target
661
def _put_on_disk(self, fullpath, tree):
662
"""See InventoryEntry._put_on_disk."""
664
os.symlink(self.symlink_target, fullpath)
666
raise BzrError("Failed to create symlink %r -> %r, error: %s" % (fullpath, self.symlink_target, e))
668
def _read_tree_state(self, path, work_tree):
669
"""See InventoryEntry._read_tree_state."""
670
self.symlink_target = work_tree.get_symlink_target(self.file_id)
672
def _forget_tree_state(self):
673
self.symlink_target = None
675
def _unchanged(self, previous_ie):
676
"""See InventoryEntry._unchanged."""
677
compatible = super(InventoryLink, self)._unchanged(previous_ie)
678
if self.symlink_target != previous_ie.symlink_target:
683
class TreeReference(InventoryEntry):
685
kind = 'tree-reference'
687
def __init__(self, file_id, name, parent_id, revision=None,
688
reference_revision=None):
689
InventoryEntry.__init__(self, file_id, name, parent_id)
690
self.revision = revision
691
self.reference_revision = reference_revision
694
return TreeReference(self.file_id, self.name, self.parent_id,
695
self.revision, self.reference_revision)
697
def _read_tree_state(self, path, work_tree):
698
"""Populate fields in the inventory entry from the given tree.
700
self.reference_revision = work_tree.get_reference_revision(
703
def _forget_tree_state(self):
704
self.reference_revision = None
706
def _unchanged(self, previous_ie):
707
"""See InventoryEntry._unchanged."""
708
compatible = super(TreeReference, self)._unchanged(previous_ie)
709
if self.reference_revision != previous_ie.reference_revision:
714
class CommonInventory(object):
715
"""Basic inventory logic, defined in terms of primitives like has_id.
717
An inventory is the metadata about the contents of a tree.
719
This is broadly a map from file_id to entries such as directories, files,
720
symlinks and tree references. Each entry maintains its own metadata like
721
SHA1 and length for files, or children for a directory.
723
Entries can be looked up either by path or by file_id.
725
InventoryEntry objects must not be modified after they are
726
inserted, other than through the Inventory API.
729
def __contains__(self, file_id):
730
"""True if this entry contains a file with given id.
732
>>> inv = Inventory()
733
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
734
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
740
Note that this method along with __iter__ are not encouraged for use as
741
they are less clear than specific query methods - they may be rmeoved
744
return self.has_id(file_id)
746
def id2path(self, file_id):
747
"""Return as a string the path to file_id.
750
>>> e = i.add(InventoryDirectory('src-id', 'src', ROOT_ID))
751
>>> e = i.add(InventoryFile('foo-id', 'foo.c', parent_id='src-id'))
752
>>> print i.id2path('foo-id')
755
:raises NoSuchId: If file_id is not present in the inventory.
757
# get all names, skipping root
758
return '/'.join(reversed(
759
[parent.name for parent in
760
self._iter_file_id_parents(file_id)][:-1]))
762
def iter_entries(self, from_dir=None, recursive=True):
763
"""Return (path, entry) pairs, in order by name.
765
:param from_dir: if None, start from the root,
766
otherwise start from this directory (either file-id or entry)
767
:param recursive: recurse into directories or not
770
if self.root is None:
774
elif isinstance(from_dir, basestring):
775
from_dir = self[from_dir]
777
# unrolling the recursive called changed the time from
778
# 440ms/663ms (inline/total) to 116ms/116ms
779
children = from_dir.children.items()
782
for name, ie in children:
785
children = collections.deque(children)
786
stack = [(u'', children)]
788
from_dir_relpath, children = stack[-1]
791
name, ie = children.popleft()
793
# we know that from_dir_relpath never ends in a slash
794
# and 'f' doesn't begin with one, we can do a string op, rather
795
# than the checks of pathjoin(), though this means that all paths
797
path = from_dir_relpath + '/' + name
801
if ie.kind != 'directory':
804
# But do this child first
805
new_children = ie.children.items()
807
new_children = collections.deque(new_children)
808
stack.append((path, new_children))
809
# Break out of inner loop, so that we start outer loop with child
812
# if we finished all children, pop it off the stack
815
def iter_entries_by_dir(self, from_dir=None, specific_file_ids=None,
816
yield_parents=False):
817
"""Iterate over the entries in a directory first order.
819
This returns all entries for a directory before returning
820
the entries for children of a directory. This is not
821
lexicographically sorted order, and is a hybrid between
822
depth-first and breadth-first.
824
:param yield_parents: If True, yield the parents from the root leading
825
down to specific_file_ids that have been requested. This has no
826
impact if specific_file_ids is None.
827
:return: This yields (path, entry) pairs
829
if specific_file_ids and not isinstance(specific_file_ids, set):
830
specific_file_ids = set(specific_file_ids)
831
# TODO? Perhaps this should return the from_dir so that the root is
832
# yielded? or maybe an option?
834
if self.root is None:
836
# Optimize a common case
837
if (not yield_parents and specific_file_ids is not None and
838
len(specific_file_ids) == 1):
839
file_id = list(specific_file_ids)[0]
841
yield self.id2path(file_id), self[file_id]
844
if (specific_file_ids is None or yield_parents or
845
self.root.file_id in specific_file_ids):
847
elif isinstance(from_dir, basestring):
848
from_dir = self[from_dir]
850
if specific_file_ids is not None:
851
# TODO: jam 20070302 This could really be done as a loop rather
852
# than a bunch of recursive calls.
855
def add_ancestors(file_id):
856
if file_id not in byid:
858
parent_id = byid[file_id].parent_id
859
if parent_id is None:
861
if parent_id not in parents:
862
parents.add(parent_id)
863
add_ancestors(parent_id)
864
for file_id in specific_file_ids:
865
add_ancestors(file_id)
869
stack = [(u'', from_dir)]
871
cur_relpath, cur_dir = stack.pop()
874
for child_name, child_ie in sorted(cur_dir.children.iteritems()):
876
child_relpath = cur_relpath + child_name
878
if (specific_file_ids is None or
879
child_ie.file_id in specific_file_ids or
880
(yield_parents and child_ie.file_id in parents)):
881
yield child_relpath, child_ie
883
if child_ie.kind == 'directory':
884
if parents is None or child_ie.file_id in parents:
885
child_dirs.append((child_relpath+'/', child_ie))
886
stack.extend(reversed(child_dirs))
888
def _make_delta(self, old):
889
"""Make an inventory delta from two inventories."""
892
adds = new_ids - old_ids
893
deletes = old_ids - new_ids
894
common = old_ids.intersection(new_ids)
896
for file_id in deletes:
897
delta.append((old.id2path(file_id), None, file_id, None))
899
delta.append((None, self.id2path(file_id), file_id, self[file_id]))
900
for file_id in common:
901
if old[file_id] != self[file_id]:
902
delta.append((old.id2path(file_id), self.id2path(file_id),
903
file_id, self[file_id]))
906
def _get_mutable_inventory(self):
907
"""Returns a mutable copy of the object.
909
Some inventories are immutable, yet working trees, for example, needs
910
to mutate exisiting inventories instead of creating a new one.
912
raise NotImplementedError(self._get_mutable_inventory)
914
def make_entry(self, kind, name, parent_id, file_id=None):
915
"""Simple thunk to bzrlib.inventory.make_entry."""
916
return make_entry(kind, name, parent_id, file_id)
919
"""Return list of (path, ie) for all entries except the root.
921
This may be faster than iter_entries.
924
def descend(dir_ie, dir_path):
925
kids = dir_ie.children.items()
927
for name, ie in kids:
928
child_path = osutils.pathjoin(dir_path, name)
929
accum.append((child_path, ie))
930
if ie.kind == 'directory':
931
descend(ie, child_path)
933
descend(self.root, u'')
936
def directories(self):
937
"""Return (path, entry) pairs for all directories, including the root.
940
def descend(parent_ie, parent_path):
941
accum.append((parent_path, parent_ie))
943
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
946
for name, child_ie in kids:
947
child_path = osutils.pathjoin(parent_path, name)
948
descend(child_ie, child_path)
949
descend(self.root, u'')
952
def path2id(self, name):
953
"""Walk down through directories to return entry of last component.
955
names may be either a list of path components, or a single
956
string, in which case it is automatically split.
958
This returns the entry of the last component in the path,
959
which may be either a file or a directory.
961
Returns None IFF the path is not found.
963
if isinstance(name, basestring):
964
name = osutils.splitpath(name)
966
# mutter("lookup path %r" % name)
970
except errors.NoSuchId:
971
# root doesn't exist yet so nothing else can
977
children = getattr(parent, 'children', None)
986
return parent.file_id
988
def filter(self, specific_fileids):
989
"""Get an inventory view filtered against a set of file-ids.
991
Children of directories and parents are included.
993
The result may or may not reference the underlying inventory
994
so it should be treated as immutable.
996
interesting_parents = set()
997
for fileid in specific_fileids:
999
interesting_parents.update(self.get_idpath(fileid))
1000
except errors.NoSuchId:
1001
# This fileid is not in the inventory - that's ok
1003
entries = self.iter_entries()
1004
if self.root is None:
1005
return Inventory(root_id=None)
1006
other = Inventory(entries.next()[1].file_id)
1007
other.root.revision = self.root.revision
1008
other.revision_id = self.revision_id
1009
directories_to_expand = set()
1010
for path, entry in entries:
1011
file_id = entry.file_id
1012
if (file_id in specific_fileids
1013
or entry.parent_id in directories_to_expand):
1014
if entry.kind == 'directory':
1015
directories_to_expand.add(file_id)
1016
elif file_id not in interesting_parents:
1018
other.add(entry.copy())
1021
def get_idpath(self, file_id):
1022
"""Return a list of file_ids for the path to an entry.
1024
The list contains one element for each directory followed by
1025
the id of the file itself. So the length of the returned list
1026
is equal to the depth of the file in the tree, counting the
1027
root directory as depth 1.
1030
for parent in self._iter_file_id_parents(file_id):
1031
p.insert(0, parent.file_id)
1035
class Inventory(CommonInventory):
1036
"""Mutable dict based in-memory inventory.
1038
We never store the full path to a file, because renaming a directory
1039
implicitly moves all of its contents. This class internally maintains a
1040
lookup tree that allows the children under a directory to be
1043
>>> inv = Inventory()
1044
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
1045
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
1046
>>> inv['123-123'].name
1049
Id's may be looked up from paths:
1051
>>> inv.path2id('hello.c')
1053
>>> '123-123' in inv
1056
There are iterators over the contents:
1058
>>> [entry[0] for entry in inv.iter_entries()]
1062
def __init__(self, root_id=ROOT_ID, revision_id=None):
1063
"""Create or read an inventory.
1065
If a working directory is specified, the inventory is read
1066
from there. If the file is specified, read from that. If not,
1067
the inventory is created empty.
1069
The inventory is created with a default root directory, with
1072
if root_id is not None:
1073
self._set_root(InventoryDirectory(root_id, u'', None))
1077
self.revision_id = revision_id
1080
# More than one page of ouput is not useful anymore to debug
1083
contents = repr(self._byid)
1084
if len(contents) > max_len:
1085
contents = contents[:(max_len-len(closing))] + closing
1086
return "<Inventory object at %x, contents=%r>" % (id(self), contents)
1088
def apply_delta(self, delta):
1089
"""Apply a delta to this inventory.
1091
See the inventory developers documentation for the theory behind
1094
If delta application fails the inventory is left in an indeterminate
1095
state and must not be used.
1097
:param delta: A list of changes to apply. After all the changes are
1098
applied the final inventory must be internally consistent, but it
1099
is ok to supply changes which, if only half-applied would have an
1100
invalid result - such as supplying two changes which rename two
1101
files, 'A' and 'B' with each other : [('A', 'B', 'A-id', a_entry),
1102
('B', 'A', 'B-id', b_entry)].
1104
Each change is a tuple, of the form (old_path, new_path, file_id,
1107
When new_path is None, the change indicates the removal of an entry
1108
from the inventory and new_entry will be ignored (using None is
1109
appropriate). If new_path is not None, then new_entry must be an
1110
InventoryEntry instance, which will be incorporated into the
1111
inventory (and replace any existing entry with the same file id).
1113
When old_path is None, the change indicates the addition of
1114
a new entry to the inventory.
1116
When neither new_path nor old_path are None, the change is a
1117
modification to an entry, such as a rename, reparent, kind change
1120
The children attribute of new_entry is ignored. This is because
1121
this method preserves children automatically across alterations to
1122
the parent of the children, and cases where the parent id of a
1123
child is changing require the child to be passed in as a separate
1124
change regardless. E.g. in the recursive deletion of a directory -
1125
the directory's children must be included in the delta, or the
1126
final inventory will be invalid.
1128
Note that a file_id must only appear once within a given delta.
1129
An AssertionError is raised otherwise.
1131
# Check that the delta is legal. It would be nice if this could be
1132
# done within the loops below but it's safer to validate the delta
1133
# before starting to mutate the inventory, as there isn't a rollback
1135
list(_check_delta_unique_ids(_check_delta_unique_new_paths(
1136
_check_delta_unique_old_paths(_check_delta_ids_match_entry(
1137
_check_delta_ids_are_valid(
1138
_check_delta_new_path_entry_both_or_None(
1142
# Remove all affected items which were in the original inventory,
1143
# starting with the longest paths, thus ensuring parents are examined
1144
# after their children, which means that everything we examine has no
1145
# modified children remaining by the time we examine it.
1146
for old_path, file_id in sorted(((op, f) for op, np, f, e in delta
1147
if op is not None), reverse=True):
1148
# Preserve unaltered children of file_id for later reinsertion.
1149
file_id_children = getattr(self[file_id], 'children', {})
1150
if len(file_id_children):
1151
children[file_id] = file_id_children
1152
if self.id2path(file_id) != old_path:
1153
raise errors.InconsistentDelta(old_path, file_id,
1154
"Entry was at wrong other path %r." % self.id2path(file_id))
1155
# Remove file_id and the unaltered children. If file_id is not
1156
# being deleted it will be reinserted back later.
1157
self.remove_recursive_id(file_id)
1158
# Insert all affected which should be in the new inventory, reattaching
1159
# their children if they had any. This is done from shortest path to
1160
# longest, ensuring that items which were modified and whose parents in
1161
# the resulting inventory were also modified, are inserted after their
1163
for new_path, f, new_entry in sorted((np, f, e) for op, np, f, e in
1164
delta if np is not None):
1165
if new_entry.kind == 'directory':
1166
# Pop the child which to allow detection of children whose
1167
# parents were deleted and which were not reattached to a new
1169
replacement = InventoryDirectory(new_entry.file_id,
1170
new_entry.name, new_entry.parent_id)
1171
replacement.revision = new_entry.revision
1172
replacement.children = children.pop(replacement.file_id, {})
1173
new_entry = replacement
1176
except errors.DuplicateFileId:
1177
raise errors.InconsistentDelta(new_path, new_entry.file_id,
1178
"New id is already present in target.")
1179
except AttributeError:
1180
raise errors.InconsistentDelta(new_path, new_entry.file_id,
1181
"Parent is not a directory.")
1182
if self.id2path(new_entry.file_id) != new_path:
1183
raise errors.InconsistentDelta(new_path, new_entry.file_id,
1184
"New path is not consistent with parent path.")
1186
# Get the parent id that was deleted
1187
parent_id, children = children.popitem()
1188
raise errors.InconsistentDelta("<deleted>", parent_id,
1189
"The file id was deleted but its children were not deleted.")
1191
def _set_root(self, ie):
1193
self._byid = {self.root.file_id: self.root}
1196
# TODO: jam 20051218 Should copy also copy the revision_id?
1197
entries = self.iter_entries()
1198
if self.root is None:
1199
return Inventory(root_id=None)
1200
other = Inventory(entries.next()[1].file_id)
1201
other.root.revision = self.root.revision
1202
# copy recursively so we know directories will be added before
1203
# their children. There are more efficient ways than this...
1204
for path, entry in entries:
1205
other.add(entry.copy())
1208
def _get_mutable_inventory(self):
1209
"""See CommonInventory._get_mutable_inventory."""
1210
return copy.deepcopy(self)
1213
"""Iterate over all file-ids."""
1214
return iter(self._byid)
1216
def iter_just_entries(self):
1217
"""Iterate over all entries.
1219
Unlike iter_entries(), just the entries are returned (not (path, ie))
1220
and the order of entries is undefined.
1222
XXX: We may not want to merge this into bzr.dev.
1224
if self.root is None:
1226
for _, ie in self._byid.iteritems():
1230
"""Returns number of entries."""
1231
return len(self._byid)
1233
def __getitem__(self, file_id):
1234
"""Return the entry for given file_id.
1236
>>> inv = Inventory()
1237
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
1238
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
1239
>>> inv['123123'].name
1243
return self._byid[file_id]
1245
# really we're passing an inventory, not a tree...
1246
raise errors.NoSuchId(self, file_id)
1248
def get_file_kind(self, file_id):
1249
return self._byid[file_id].kind
1251
def get_child(self, parent_id, filename):
1252
return self[parent_id].children.get(filename)
1254
def _add_child(self, entry):
1255
"""Add an entry to the inventory, without adding it to its parent"""
1256
if entry.file_id in self._byid:
1257
raise BzrError("inventory already contains entry with id {%s}" %
1259
self._byid[entry.file_id] = entry
1260
for child in getattr(entry, 'children', {}).itervalues():
1261
self._add_child(child)
1264
def add(self, entry):
1265
"""Add entry to inventory.
1267
To add a file to a branch ready to be committed, use Branch.add,
1272
if entry.file_id in self._byid:
1273
raise errors.DuplicateFileId(entry.file_id,
1274
self._byid[entry.file_id])
1275
if entry.parent_id is None:
1279
parent = self._byid[entry.parent_id]
1281
raise errors.InconsistentDelta("<unknown>", entry.parent_id,
1282
"Parent not in inventory.")
1283
if entry.name in parent.children:
1284
raise errors.InconsistentDelta(
1285
self.id2path(parent.children[entry.name].file_id),
1287
"Path already versioned")
1288
parent.children[entry.name] = entry
1289
return self._add_child(entry)
1291
def add_path(self, relpath, kind, file_id=None, parent_id=None):
1292
"""Add entry from a path.
1294
The immediate parent must already be versioned.
1296
Returns the new entry object."""
1298
parts = osutils.splitpath(relpath)
1302
file_id = generate_ids.gen_root_id()
1303
self.root = InventoryDirectory(file_id, '', None)
1304
self._byid = {self.root.file_id: self.root}
1307
parent_path = parts[:-1]
1308
parent_id = self.path2id(parent_path)
1309
if parent_id is None:
1310
raise errors.NotVersionedError(path=parent_path)
1311
ie = make_entry(kind, parts[-1], parent_id, file_id)
1314
def __delitem__(self, file_id):
1315
"""Remove entry by id.
1317
>>> inv = Inventory()
1318
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1319
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
1327
del self._byid[file_id]
1328
if ie.parent_id is not None:
1329
del self[ie.parent_id].children[ie.name]
1331
def __eq__(self, other):
1332
"""Compare two sets by comparing their contents.
1334
>>> i1 = Inventory()
1335
>>> i2 = Inventory()
1338
>>> i1.add(InventoryFile('123', 'foo', ROOT_ID))
1339
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
1342
>>> i2.add(InventoryFile('123', 'foo', ROOT_ID))
1343
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
1347
if not isinstance(other, Inventory):
1348
return NotImplemented
1350
return self._byid == other._byid
1352
def __ne__(self, other):
1353
return not self.__eq__(other)
1356
raise ValueError('not hashable')
1358
def _iter_file_id_parents(self, file_id):
1359
"""Yield the parents of file_id up to the root."""
1360
while file_id is not None:
1362
ie = self._byid[file_id]
1364
raise errors.NoSuchId(tree=None, file_id=file_id)
1366
file_id = ie.parent_id
1368
def has_filename(self, filename):
1369
return bool(self.path2id(filename))
1371
def has_id(self, file_id):
1372
return (file_id in self._byid)
1374
def _make_delta(self, old):
1375
"""Make an inventory delta from two inventories."""
1376
old_getter = getattr(old, '_byid', old)
1377
new_getter = self._byid
1378
old_ids = set(old_getter)
1379
new_ids = set(new_getter)
1380
adds = new_ids - old_ids
1381
deletes = old_ids - new_ids
1382
if not adds and not deletes:
1385
common = old_ids.intersection(new_ids)
1387
for file_id in deletes:
1388
delta.append((old.id2path(file_id), None, file_id, None))
1389
for file_id in adds:
1390
delta.append((None, self.id2path(file_id), file_id, self[file_id]))
1391
for file_id in common:
1392
new_ie = new_getter[file_id]
1393
old_ie = old_getter[file_id]
1394
# If xml_serializer returns the cached InventoryEntries (rather
1395
# than always doing .copy()), inlining the 'is' check saves 2.7M
1396
# calls to __eq__. Under lsprof this saves 20s => 6s.
1397
# It is a minor improvement without lsprof.
1398
if old_ie is new_ie or old_ie == new_ie:
1401
delta.append((old.id2path(file_id), self.id2path(file_id),
1405
def remove_recursive_id(self, file_id):
1406
"""Remove file_id, and children, from the inventory.
1408
:param file_id: A file_id to remove.
1410
to_find_delete = [self._byid[file_id]]
1412
while to_find_delete:
1413
ie = to_find_delete.pop()
1414
to_delete.append(ie.file_id)
1415
if ie.kind == 'directory':
1416
to_find_delete.extend(ie.children.values())
1417
for file_id in reversed(to_delete):
1419
del self._byid[file_id]
1420
if ie.parent_id is not None:
1421
del self[ie.parent_id].children[ie.name]
1425
def rename(self, file_id, new_parent_id, new_name):
1426
"""Move a file within the inventory.
1428
This can change either the name, or the parent, or both.
1430
This does not move the working file.
1432
new_name = ensure_normalized_name(new_name)
1433
if not is_valid_name(new_name):
1434
raise BzrError("not an acceptable filename: %r" % new_name)
1436
new_parent = self._byid[new_parent_id]
1437
if new_name in new_parent.children:
1438
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
1440
new_parent_idpath = self.get_idpath(new_parent_id)
1441
if file_id in new_parent_idpath:
1442
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
1443
% (self.id2path(file_id), self.id2path(new_parent_id)))
1445
file_ie = self._byid[file_id]
1446
old_parent = self._byid[file_ie.parent_id]
1448
# TODO: Don't leave things messed up if this fails
1450
del old_parent.children[file_ie.name]
1451
new_parent.children[new_name] = file_ie
1453
file_ie.name = new_name
1454
file_ie.parent_id = new_parent_id
1456
def is_root(self, file_id):
1457
return self.root is not None and file_id == self.root.file_id
1460
class CHKInventory(CommonInventory):
1461
"""An inventory persisted in a CHK store.
1463
By design, a CHKInventory is immutable so many of the methods
1464
supported by Inventory - add, rename, apply_delta, etc - are *not*
1465
supported. To create a new CHKInventory, use create_by_apply_delta()
1466
or from_inventory(), say.
1468
Internally, a CHKInventory has one or two CHKMaps:
1470
* id_to_entry - a map from (file_id,) => InventoryEntry as bytes
1471
* parent_id_basename_to_file_id - a map from (parent_id, basename_utf8)
1474
The second map is optional and not present in early CHkRepository's.
1476
No caching is performed: every method call or item access will perform
1477
requests to the storage layer. As such, keep references to objects you
1481
def __init__(self, search_key_name):
1482
CommonInventory.__init__(self)
1483
self._fileid_to_entry_cache = {}
1484
self._path_to_fileid_cache = {}
1485
self._search_key_name = search_key_name
1488
def __eq__(self, other):
1489
"""Compare two sets by comparing their contents."""
1490
if not isinstance(other, CHKInventory):
1491
return NotImplemented
1493
this_key = self.id_to_entry.key()
1494
other_key = other.id_to_entry.key()
1495
this_pid_key = self.parent_id_basename_to_file_id.key()
1496
other_pid_key = other.parent_id_basename_to_file_id.key()
1497
if None in (this_key, this_pid_key, other_key, other_pid_key):
1499
return this_key == other_key and this_pid_key == other_pid_key
1501
def _entry_to_bytes(self, entry):
1502
"""Serialise entry as a single bytestring.
1504
:param Entry: An inventory entry.
1505
:return: A bytestring for the entry.
1508
ENTRY ::= FILE | DIR | SYMLINK | TREE
1509
FILE ::= "file: " COMMON SEP SHA SEP SIZE SEP EXECUTABLE
1510
DIR ::= "dir: " COMMON
1511
SYMLINK ::= "symlink: " COMMON SEP TARGET_UTF8
1512
TREE ::= "tree: " COMMON REFERENCE_REVISION
1513
COMMON ::= FILE_ID SEP PARENT_ID SEP NAME_UTF8 SEP REVISION
1516
if entry.parent_id is not None:
1517
parent_str = entry.parent_id
1520
name_str = entry.name.encode("utf8")
1521
if entry.kind == 'file':
1522
if entry.executable:
1526
return "file: %s\n%s\n%s\n%s\n%s\n%d\n%s" % (
1527
entry.file_id, parent_str, name_str, entry.revision,
1528
entry.text_sha1, entry.text_size, exec_str)
1529
elif entry.kind == 'directory':
1530
return "dir: %s\n%s\n%s\n%s" % (
1531
entry.file_id, parent_str, name_str, entry.revision)
1532
elif entry.kind == 'symlink':
1533
return "symlink: %s\n%s\n%s\n%s\n%s" % (
1534
entry.file_id, parent_str, name_str, entry.revision,
1535
entry.symlink_target.encode("utf8"))
1536
elif entry.kind == 'tree-reference':
1537
return "tree: %s\n%s\n%s\n%s\n%s" % (
1538
entry.file_id, parent_str, name_str, entry.revision,
1539
entry.reference_revision)
1541
raise ValueError("unknown kind %r" % entry.kind)
1544
def _bytes_to_utf8name_key(bytes):
1545
"""Get the file_id, revision_id key out of bytes."""
1546
# We don't normally care about name, except for times when we want
1547
# to filter out empty names because of non rich-root...
1548
sections = bytes.split('\n')
1549
kind, file_id = sections[0].split(': ')
1550
return (sections[2], file_id, sections[3])
1552
def _bytes_to_entry(self, bytes):
1553
"""Deserialise a serialised entry."""
1554
sections = bytes.split('\n')
1555
if sections[0].startswith("file: "):
1556
result = InventoryFile(sections[0][6:],
1557
sections[2].decode('utf8'),
1559
result.text_sha1 = sections[4]
1560
result.text_size = int(sections[5])
1561
result.executable = sections[6] == "Y"
1562
elif sections[0].startswith("dir: "):
1563
result = CHKInventoryDirectory(sections[0][5:],
1564
sections[2].decode('utf8'),
1566
elif sections[0].startswith("symlink: "):
1567
result = InventoryLink(sections[0][9:],
1568
sections[2].decode('utf8'),
1570
result.symlink_target = sections[4].decode('utf8')
1571
elif sections[0].startswith("tree: "):
1572
result = TreeReference(sections[0][6:],
1573
sections[2].decode('utf8'),
1575
result.reference_revision = sections[4]
1577
raise ValueError("Not a serialised entry %r" % bytes)
1578
result.revision = sections[3]
1579
if result.parent_id == '':
1580
result.parent_id = None
1581
self._fileid_to_entry_cache[result.file_id] = result
1584
def _get_mutable_inventory(self):
1585
"""See CommonInventory._get_mutable_inventory."""
1586
entries = self.iter_entries()
1587
inv = Inventory(None, self.revision_id)
1588
for path, inv_entry in entries:
1589
inv.add(inv_entry.copy())
1592
def create_by_apply_delta(self, inventory_delta, new_revision_id,
1593
propagate_caches=False):
1594
"""Create a new CHKInventory by applying inventory_delta to this one.
1596
See the inventory developers documentation for the theory behind
1599
:param inventory_delta: The inventory delta to apply. See
1600
Inventory.apply_delta for details.
1601
:param new_revision_id: The revision id of the resulting CHKInventory.
1602
:param propagate_caches: If True, the caches for this inventory are
1603
copied to and updated for the result.
1604
:return: The new CHKInventory.
1606
split = osutils.split
1607
result = CHKInventory(self._search_key_name)
1608
if propagate_caches:
1609
# Just propagate the path-to-fileid cache for now
1610
result._path_to_fileid_cache = dict(self._path_to_fileid_cache.iteritems())
1611
search_key_func = chk_map.search_key_registry.get(self._search_key_name)
1612
self.id_to_entry._ensure_root()
1613
maximum_size = self.id_to_entry._root_node.maximum_size
1614
result.revision_id = new_revision_id
1615
result.id_to_entry = chk_map.CHKMap(
1616
self.id_to_entry._store,
1617
self.id_to_entry.key(),
1618
search_key_func=search_key_func)
1619
result.id_to_entry._ensure_root()
1620
result.id_to_entry._root_node.set_maximum_size(maximum_size)
1621
# Change to apply to the parent_id_basename delta. The dict maps
1622
# (parent_id, basename) -> (old_key, new_value). We use a dict because
1623
# when a path has its id replaced (e.g. the root is changed, or someone
1624
# does bzr mv a b, bzr mv c a, we should output a single change to this
1625
# map rather than two.
1626
parent_id_basename_delta = {}
1627
if self.parent_id_basename_to_file_id is not None:
1628
result.parent_id_basename_to_file_id = chk_map.CHKMap(
1629
self.parent_id_basename_to_file_id._store,
1630
self.parent_id_basename_to_file_id.key(),
1631
search_key_func=search_key_func)
1632
result.parent_id_basename_to_file_id._ensure_root()
1633
self.parent_id_basename_to_file_id._ensure_root()
1634
result_p_id_root = result.parent_id_basename_to_file_id._root_node
1635
p_id_root = self.parent_id_basename_to_file_id._root_node
1636
result_p_id_root.set_maximum_size(p_id_root.maximum_size)
1637
result_p_id_root._key_width = p_id_root._key_width
1639
result.parent_id_basename_to_file_id = None
1640
result.root_id = self.root_id
1641
id_to_entry_delta = []
1642
# inventory_delta is only traversed once, so we just update the
1644
# Check for repeated file ids
1645
inventory_delta = _check_delta_unique_ids(inventory_delta)
1646
# Repeated old paths
1647
inventory_delta = _check_delta_unique_old_paths(inventory_delta)
1648
# Check for repeated new paths
1649
inventory_delta = _check_delta_unique_new_paths(inventory_delta)
1650
# Check for entries that don't match the fileid
1651
inventory_delta = _check_delta_ids_match_entry(inventory_delta)
1652
# Check for nonsense fileids
1653
inventory_delta = _check_delta_ids_are_valid(inventory_delta)
1654
# Check for new_path <-> entry consistency
1655
inventory_delta = _check_delta_new_path_entry_both_or_None(
1657
# All changed entries need to have their parents be directories and be
1658
# at the right path. This set contains (path, id) tuples.
1660
# When we delete an item, all the children of it must be either deleted
1661
# or altered in their own right. As we batch process the change via
1662
# CHKMap.apply_delta, we build a set of things to use to validate the
1666
for old_path, new_path, file_id, entry in inventory_delta:
1669
result.root_id = file_id
1670
if new_path is None:
1675
if propagate_caches:
1677
del result._path_to_fileid_cache[old_path]
1680
deletes.add(file_id)
1682
new_key = (file_id,)
1683
new_value = result._entry_to_bytes(entry)
1684
# Update caches. It's worth doing this whether
1685
# we're propagating the old caches or not.
1686
result._path_to_fileid_cache[new_path] = file_id
1687
parents.add((split(new_path)[0], entry.parent_id))
1688
if old_path is None:
1691
old_key = (file_id,)
1692
if self.id2path(file_id) != old_path:
1693
raise errors.InconsistentDelta(old_path, file_id,
1694
"Entry was at wrong other path %r." %
1695
self.id2path(file_id))
1696
altered.add(file_id)
1697
id_to_entry_delta.append((old_key, new_key, new_value))
1698
if result.parent_id_basename_to_file_id is not None:
1699
# parent_id, basename changes
1700
if old_path is None:
1703
old_entry = self[file_id]
1704
old_key = self._parent_id_basename_key(old_entry)
1705
if new_path is None:
1709
new_key = self._parent_id_basename_key(entry)
1711
# If the two keys are the same, the value will be unchanged
1712
# as its always the file id for this entry.
1713
if old_key != new_key:
1714
# Transform a change into explicit delete/add preserving
1715
# a possible match on the key from a different file id.
1716
if old_key is not None:
1717
parent_id_basename_delta.setdefault(
1718
old_key, [None, None])[0] = old_key
1719
if new_key is not None:
1720
parent_id_basename_delta.setdefault(
1721
new_key, [None, None])[1] = new_value
1722
# validate that deletes are complete.
1723
for file_id in deletes:
1724
entry = self[file_id]
1725
if entry.kind != 'directory':
1727
# This loop could potentially be better by using the id_basename
1728
# map to just get the child file ids.
1729
for child in entry.children.values():
1730
if child.file_id not in altered:
1731
raise errors.InconsistentDelta(self.id2path(child.file_id),
1732
child.file_id, "Child not deleted or reparented when "
1734
result.id_to_entry.apply_delta(id_to_entry_delta)
1735
if parent_id_basename_delta:
1736
# Transform the parent_id_basename delta data into a linear delta
1737
# with only one record for a given key. Optimally this would allow
1738
# re-keying, but its simpler to just output that as a delete+add
1739
# to spend less time calculating the delta.
1741
for key, (old_key, value) in parent_id_basename_delta.iteritems():
1742
if value is not None:
1743
delta_list.append((old_key, key, value))
1745
delta_list.append((old_key, None, None))
1746
result.parent_id_basename_to_file_id.apply_delta(delta_list)
1747
parents.discard(('', None))
1748
for parent_path, parent in parents:
1750
if result[parent].kind != 'directory':
1751
raise errors.InconsistentDelta(result.id2path(parent), parent,
1752
'Not a directory, but given children')
1753
except errors.NoSuchId:
1754
raise errors.InconsistentDelta("<unknown>", parent,
1755
"Parent is not present in resulting inventory.")
1756
if result.path2id(parent_path) != parent:
1757
raise errors.InconsistentDelta(parent_path, parent,
1758
"Parent has wrong path %r." % result.path2id(parent_path))
1762
def deserialise(klass, chk_store, bytes, expected_revision_id):
1763
"""Deserialise a CHKInventory.
1765
:param chk_store: A CHK capable VersionedFiles instance.
1766
:param bytes: The serialised bytes.
1767
:param expected_revision_id: The revision ID we think this inventory is
1769
:return: A CHKInventory
1771
lines = bytes.split('\n')
1773
raise AssertionError('bytes to deserialize must end with an eol')
1775
if lines[0] != 'chkinventory:':
1776
raise ValueError("not a serialised CHKInventory: %r" % bytes)
1778
allowed_keys = frozenset(['root_id', 'revision_id', 'search_key_name',
1779
'parent_id_basename_to_file_id',
1781
for line in lines[1:]:
1782
key, value = line.split(': ', 1)
1783
if key not in allowed_keys:
1784
raise errors.BzrError('Unknown key in inventory: %r\n%r'
1787
raise errors.BzrError('Duplicate key in inventory: %r\n%r'
1790
revision_id = info['revision_id']
1791
root_id = info['root_id']
1792
search_key_name = info.get('search_key_name', 'plain')
1793
parent_id_basename_to_file_id = info.get(
1794
'parent_id_basename_to_file_id', None)
1795
id_to_entry = info['id_to_entry']
1797
result = CHKInventory(search_key_name)
1798
result.revision_id = revision_id
1799
result.root_id = root_id
1800
search_key_func = chk_map.search_key_registry.get(
1801
result._search_key_name)
1802
if parent_id_basename_to_file_id is not None:
1803
result.parent_id_basename_to_file_id = chk_map.CHKMap(
1804
chk_store, (parent_id_basename_to_file_id,),
1805
search_key_func=search_key_func)
1807
result.parent_id_basename_to_file_id = None
1809
result.id_to_entry = chk_map.CHKMap(chk_store, (id_to_entry,),
1810
search_key_func=search_key_func)
1811
if (result.revision_id,) != expected_revision_id:
1812
raise ValueError("Mismatched revision id and expected: %r, %r" %
1813
(result.revision_id, expected_revision_id))
1817
def from_inventory(klass, chk_store, inventory, maximum_size=0, search_key_name='plain'):
1818
"""Create a CHKInventory from an existing inventory.
1820
The content of inventory is copied into the chk_store, and a
1821
CHKInventory referencing that is returned.
1823
:param chk_store: A CHK capable VersionedFiles instance.
1824
:param inventory: The inventory to copy.
1825
:param maximum_size: The CHKMap node size limit.
1826
:param search_key_name: The identifier for the search key function
1828
result = klass(search_key_name)
1829
result.revision_id = inventory.revision_id
1830
result.root_id = inventory.root.file_id
1832
entry_to_bytes = result._entry_to_bytes
1833
parent_id_basename_key = result._parent_id_basename_key
1834
id_to_entry_dict = {}
1835
parent_id_basename_dict = {}
1836
for path, entry in inventory.iter_entries():
1837
id_to_entry_dict[(entry.file_id,)] = entry_to_bytes(entry)
1838
p_id_key = parent_id_basename_key(entry)
1839
parent_id_basename_dict[p_id_key] = entry.file_id
1841
result._populate_from_dicts(chk_store, id_to_entry_dict,
1842
parent_id_basename_dict, maximum_size=maximum_size)
1845
def _populate_from_dicts(self, chk_store, id_to_entry_dict,
1846
parent_id_basename_dict, maximum_size):
1847
search_key_func = chk_map.search_key_registry.get(self._search_key_name)
1848
root_key = chk_map.CHKMap.from_dict(chk_store, id_to_entry_dict,
1849
maximum_size=maximum_size, key_width=1,
1850
search_key_func=search_key_func)
1851
self.id_to_entry = chk_map.CHKMap(chk_store, root_key,
1853
root_key = chk_map.CHKMap.from_dict(chk_store,
1854
parent_id_basename_dict,
1855
maximum_size=maximum_size, key_width=2,
1856
search_key_func=search_key_func)
1857
self.parent_id_basename_to_file_id = chk_map.CHKMap(chk_store,
1858
root_key, search_key_func)
1860
def _parent_id_basename_key(self, entry):
1861
"""Create a key for a entry in a parent_id_basename_to_file_id index."""
1862
if entry.parent_id is not None:
1863
parent_id = entry.parent_id
1866
return parent_id, entry.name.encode('utf8')
1868
def __getitem__(self, file_id):
1869
"""map a single file_id -> InventoryEntry."""
1871
raise errors.NoSuchId(self, file_id)
1872
result = self._fileid_to_entry_cache.get(file_id, None)
1873
if result is not None:
1876
return self._bytes_to_entry(
1877
self.id_to_entry.iteritems([(file_id,)]).next()[1])
1878
except StopIteration:
1879
# really we're passing an inventory, not a tree...
1880
raise errors.NoSuchId(self, file_id)
1882
def has_id(self, file_id):
1883
# Perhaps have an explicit 'contains' method on CHKMap ?
1884
if self._fileid_to_entry_cache.get(file_id, None) is not None:
1886
return len(list(self.id_to_entry.iteritems([(file_id,)]))) == 1
1888
def is_root(self, file_id):
1889
return file_id == self.root_id
1891
def _iter_file_id_parents(self, file_id):
1892
"""Yield the parents of file_id up to the root."""
1893
while file_id is not None:
1897
raise errors.NoSuchId(tree=self, file_id=file_id)
1899
file_id = ie.parent_id
1902
"""Iterate over all file-ids."""
1903
for key, _ in self.id_to_entry.iteritems():
1906
def iter_just_entries(self):
1907
"""Iterate over all entries.
1909
Unlike iter_entries(), just the entries are returned (not (path, ie))
1910
and the order of entries is undefined.
1912
XXX: We may not want to merge this into bzr.dev.
1914
for key, entry in self.id_to_entry.iteritems():
1916
ie = self._fileid_to_entry_cache.get(file_id, None)
1918
ie = self._bytes_to_entry(entry)
1919
self._fileid_to_entry_cache[file_id] = ie
1922
def iter_changes(self, basis):
1923
"""Generate a Tree.iter_changes change list between this and basis.
1925
:param basis: Another CHKInventory.
1926
:return: An iterator over the changes between self and basis, as per
1927
tree.iter_changes().
1929
# We want: (file_id, (path_in_source, path_in_target),
1930
# changed_content, versioned, parent, name, kind,
1932
for key, basis_value, self_value in \
1933
self.id_to_entry.iter_changes(basis.id_to_entry):
1935
if basis_value is not None:
1936
basis_entry = basis._bytes_to_entry(basis_value)
1937
path_in_source = basis.id2path(file_id)
1938
basis_parent = basis_entry.parent_id
1939
basis_name = basis_entry.name
1940
basis_executable = basis_entry.executable
1942
path_in_source = None
1945
basis_executable = None
1946
if self_value is not None:
1947
self_entry = self._bytes_to_entry(self_value)
1948
path_in_target = self.id2path(file_id)
1949
self_parent = self_entry.parent_id
1950
self_name = self_entry.name
1951
self_executable = self_entry.executable
1953
path_in_target = None
1956
self_executable = None
1957
if basis_value is None:
1959
kind = (None, self_entry.kind)
1960
versioned = (False, True)
1961
elif self_value is None:
1963
kind = (basis_entry.kind, None)
1964
versioned = (True, False)
1966
kind = (basis_entry.kind, self_entry.kind)
1967
versioned = (True, True)
1968
changed_content = False
1969
if kind[0] != kind[1]:
1970
changed_content = True
1971
elif kind[0] == 'file':
1972
if (self_entry.text_size != basis_entry.text_size or
1973
self_entry.text_sha1 != basis_entry.text_sha1):
1974
changed_content = True
1975
elif kind[0] == 'symlink':
1976
if self_entry.symlink_target != basis_entry.symlink_target:
1977
changed_content = True
1978
elif kind[0] == 'tree-reference':
1979
if (self_entry.reference_revision !=
1980
basis_entry.reference_revision):
1981
changed_content = True
1982
parent = (basis_parent, self_parent)
1983
name = (basis_name, self_name)
1984
executable = (basis_executable, self_executable)
1985
if (not changed_content
1986
and parent[0] == parent[1]
1987
and name[0] == name[1]
1988
and executable[0] == executable[1]):
1989
# Could happen when only the revision changed for a directory
1992
yield (file_id, (path_in_source, path_in_target), changed_content,
1993
versioned, parent, name, kind, executable)
1996
"""Return the number of entries in the inventory."""
1997
return len(self.id_to_entry)
1999
def _make_delta(self, old):
2000
"""Make an inventory delta from two inventories."""
2001
if type(old) != CHKInventory:
2002
return CommonInventory._make_delta(self, old)
2004
for key, old_value, self_value in \
2005
self.id_to_entry.iter_changes(old.id_to_entry):
2007
if old_value is not None:
2008
old_path = old.id2path(file_id)
2011
if self_value is not None:
2012
entry = self._bytes_to_entry(self_value)
2013
self._fileid_to_entry_cache[file_id] = entry
2014
new_path = self.id2path(file_id)
2018
delta.append((old_path, new_path, file_id, entry))
2021
def path2id(self, name):
2022
"""See CommonInventory.path2id()."""
2023
# TODO: perhaps support negative hits?
2024
result = self._path_to_fileid_cache.get(name, None)
2025
if result is not None:
2027
if isinstance(name, basestring):
2028
names = osutils.splitpath(name)
2031
current_id = self.root_id
2032
if current_id is None:
2034
parent_id_index = self.parent_id_basename_to_file_id
2035
for basename in names:
2036
# TODO: Cache each path we figure out in this function.
2037
basename_utf8 = basename.encode('utf8')
2038
key_filter = [(current_id, basename_utf8)]
2040
for (parent_id, name_utf8), file_id in parent_id_index.iteritems(
2041
key_filter=key_filter):
2042
if parent_id != current_id or name_utf8 != basename_utf8:
2043
raise errors.BzrError("corrupt inventory lookup! "
2044
"%r %r %r %r" % (parent_id, current_id, name_utf8,
2048
current_id = file_id
2049
self._path_to_fileid_cache[name] = current_id
2053
"""Serialise the inventory to lines."""
2054
lines = ["chkinventory:\n"]
2055
if self._search_key_name != 'plain':
2056
# custom ordering grouping things that don't change together
2057
lines.append('search_key_name: %s\n' % (self._search_key_name,))
2058
lines.append("root_id: %s\n" % self.root_id)
2059
lines.append('parent_id_basename_to_file_id: %s\n' %
2060
self.parent_id_basename_to_file_id.key())
2061
lines.append("revision_id: %s\n" % self.revision_id)
2062
lines.append("id_to_entry: %s\n" % self.id_to_entry.key())
2064
lines.append("revision_id: %s\n" % self.revision_id)
2065
lines.append("root_id: %s\n" % self.root_id)
2066
if self.parent_id_basename_to_file_id is not None:
2067
lines.append('parent_id_basename_to_file_id: %s\n' %
2068
self.parent_id_basename_to_file_id.key())
2069
lines.append("id_to_entry: %s\n" % self.id_to_entry.key())
2074
"""Get the root entry."""
2075
return self[self.root_id]
2078
class CHKInventoryDirectory(InventoryDirectory):
2079
"""A directory in an inventory."""
2081
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
2082
'text_id', 'parent_id', '_children', 'executable',
2083
'revision', 'symlink_target', 'reference_revision',
2086
def __init__(self, file_id, name, parent_id, chk_inventory):
2087
# Don't call InventoryDirectory.__init__ - it isn't right for this
2089
InventoryEntry.__init__(self, file_id, name, parent_id)
2090
self._children = None
2091
self.kind = 'directory'
2092
self._chk_inventory = chk_inventory
2096
"""Access the list of children of this directory.
2098
With a parent_id_basename_to_file_id index, loads all the children,
2099
without loads the entire index. Without is bad. A more sophisticated
2100
proxy object might be nice, to allow partial loading of children as
2101
well when specific names are accessed. (So path traversal can be
2102
written in the obvious way but not examine siblings.).
2104
if self._children is not None:
2105
return self._children
2106
# No longer supported
2107
if self._chk_inventory.parent_id_basename_to_file_id is None:
2108
raise AssertionError("Inventories without"
2109
" parent_id_basename_to_file_id are no longer supported")
2111
# XXX: Todo - use proxy objects for the children rather than loading
2112
# all when the attribute is referenced.
2113
parent_id_index = self._chk_inventory.parent_id_basename_to_file_id
2115
for (parent_id, name_utf8), file_id in parent_id_index.iteritems(
2116
key_filter=[(self.file_id,)]):
2117
child_keys.add((file_id,))
2119
for file_id_key in child_keys:
2120
entry = self._chk_inventory._fileid_to_entry_cache.get(
2121
file_id_key[0], None)
2122
if entry is not None:
2123
result[entry.name] = entry
2124
cached.add(file_id_key)
2125
child_keys.difference_update(cached)
2126
# populate; todo: do by name
2127
id_to_entry = self._chk_inventory.id_to_entry
2128
for file_id_key, bytes in id_to_entry.iteritems(child_keys):
2129
entry = self._chk_inventory._bytes_to_entry(bytes)
2130
result[entry.name] = entry
2131
self._chk_inventory._fileid_to_entry_cache[file_id_key[0]] = entry
2132
self._children = result
2136
'directory': InventoryDirectory,
2137
'file': InventoryFile,
2138
'symlink': InventoryLink,
2139
'tree-reference': TreeReference
2142
def make_entry(kind, name, parent_id, file_id=None):
2143
"""Create an inventory entry.
2145
:param kind: the type of inventory entry to create.
2146
:param name: the basename of the entry.
2147
:param parent_id: the parent_id of the entry.
2148
:param file_id: the file_id to use. if None, one will be created.
2151
file_id = generate_ids.gen_file_id(name)
2152
name = ensure_normalized_name(name)
2154
factory = entry_factory[kind]
2156
raise BzrError("unknown kind %r" % kind)
2157
return factory(file_id, name, parent_id)
2160
def ensure_normalized_name(name):
2163
:raises InvalidNormalization: When name is not normalized, and cannot be
2164
accessed on this platform by the normalized path.
2165
:return: The NFC normalised version of name.
2167
#------- This has been copied to bzrlib.dirstate.DirState.add, please
2168
# keep them synchronised.
2169
# we dont import normalized_filename directly because we want to be
2170
# able to change the implementation at runtime for tests.
2171
norm_name, can_access = osutils.normalized_filename(name)
2172
if norm_name != name:
2176
# TODO: jam 20060701 This would probably be more useful
2177
# if the error was raised with the full path
2178
raise errors.InvalidNormalization(name)
2184
def is_valid_name(name):
2186
if _NAME_RE is None:
2187
_NAME_RE = re.compile(r'^[^/\\]+$')
2189
return bool(_NAME_RE.match(name))
2192
def _check_delta_unique_ids(delta):
2193
"""Decorate a delta and check that the file ids in it are unique.
2195
:return: A generator over delta.
2199
length = len(ids) + 1
2201
if len(ids) != length:
2202
raise errors.InconsistentDelta(item[0] or item[1], item[2],
2207
def _check_delta_unique_new_paths(delta):
2208
"""Decorate a delta and check that the new paths in it are unique.
2210
:return: A generator over delta.
2214
length = len(paths) + 1
2216
if path is not None:
2218
if len(paths) != length:
2219
raise errors.InconsistentDelta(path, item[2], "repeated path")
2223
def _check_delta_unique_old_paths(delta):
2224
"""Decorate a delta and check that the old paths in it are unique.
2226
:return: A generator over delta.
2230
length = len(paths) + 1
2232
if path is not None:
2234
if len(paths) != length:
2235
raise errors.InconsistentDelta(path, item[2], "repeated path")
2239
def _check_delta_ids_are_valid(delta):
2240
"""Decorate a delta and check that the ids in it are valid.
2242
:return: A generator over delta.
2247
raise errors.InconsistentDelta(item[0] or item[1], item[2],
2248
"entry with file_id None %r" % entry)
2249
if type(item[2]) != str:
2250
raise errors.InconsistentDelta(item[0] or item[1], item[2],
2251
"entry with non bytes file_id %r" % entry)
2255
def _check_delta_ids_match_entry(delta):
2256
"""Decorate a delta and check that the ids in it match the entry.file_id.
2258
:return: A generator over delta.
2262
if entry is not None:
2263
if entry.file_id != item[2]:
2264
raise errors.InconsistentDelta(item[0] or item[1], item[2],
2265
"mismatched id with %r" % entry)
2269
def _check_delta_new_path_entry_both_or_None(delta):
2270
"""Decorate a delta and check that the new_path and entry are paired.
2272
:return: A generator over delta.
2277
if new_path is None and entry is not None:
2278
raise errors.InconsistentDelta(item[0], item[1],
2279
"Entry with no new_path")
2280
if new_path is not None and entry is None:
2281
raise errors.InconsistentDelta(new_path, item[1],
2282
"new_path with no entry")
added
removed
bzrlib/inventory.py
