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
def __contains__(self, file_id):
718
"""True if this entry contains a file with given id.
720
>>> inv = Inventory()
721
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
722
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
728
Note that this method along with __iter__ are not encouraged for use as
729
they are less clear than specific query methods - they may be rmeoved
732
return self.has_id(file_id)
734
def id2path(self, file_id):
735
"""Return as a string the path to file_id.
738
>>> e = i.add(InventoryDirectory('src-id', 'src', ROOT_ID))
739
>>> e = i.add(InventoryFile('foo-id', 'foo.c', parent_id='src-id'))
740
>>> print i.id2path('foo-id')
743
# get all names, skipping root
744
return '/'.join(reversed(
745
[parent.name for parent in
746
self._iter_file_id_parents(file_id)][:-1]))
748
def iter_entries(self, from_dir=None, recursive=True):
749
"""Return (path, entry) pairs, in order by name.
751
:param from_dir: if None, start from the root,
752
otherwise start from this directory (either file-id or entry)
753
:param recursive: recurse into directories or not
756
if self.root is None:
760
elif isinstance(from_dir, basestring):
761
from_dir = self[from_dir]
763
# unrolling the recursive called changed the time from
764
# 440ms/663ms (inline/total) to 116ms/116ms
765
children = from_dir.children.items()
768
for name, ie in children:
771
children = collections.deque(children)
772
stack = [(u'', children)]
774
from_dir_relpath, children = stack[-1]
777
name, ie = children.popleft()
779
# we know that from_dir_relpath never ends in a slash
780
# and 'f' doesn't begin with one, we can do a string op, rather
781
# than the checks of pathjoin(), though this means that all paths
783
path = from_dir_relpath + '/' + name
787
if ie.kind != 'directory':
790
# But do this child first
791
new_children = ie.children.items()
793
new_children = collections.deque(new_children)
794
stack.append((path, new_children))
795
# Break out of inner loop, so that we start outer loop with child
798
# if we finished all children, pop it off the stack
801
def iter_entries_by_dir(self, from_dir=None, specific_file_ids=None,
802
yield_parents=False):
803
"""Iterate over the entries in a directory first order.
805
This returns all entries for a directory before returning
806
the entries for children of a directory. This is not
807
lexicographically sorted order, and is a hybrid between
808
depth-first and breadth-first.
810
:param yield_parents: If True, yield the parents from the root leading
811
down to specific_file_ids that have been requested. This has no
812
impact if specific_file_ids is None.
813
:return: This yields (path, entry) pairs
815
if specific_file_ids and not isinstance(specific_file_ids, set):
816
specific_file_ids = set(specific_file_ids)
817
# TODO? Perhaps this should return the from_dir so that the root is
818
# yielded? or maybe an option?
820
if self.root is None:
822
# Optimize a common case
823
if (not yield_parents and specific_file_ids is not None and
824
len(specific_file_ids) == 1):
825
file_id = list(specific_file_ids)[0]
827
yield self.id2path(file_id), self[file_id]
830
if (specific_file_ids is None or yield_parents or
831
self.root.file_id in specific_file_ids):
833
elif isinstance(from_dir, basestring):
834
from_dir = self[from_dir]
836
if specific_file_ids is not None:
837
# TODO: jam 20070302 This could really be done as a loop rather
838
# than a bunch of recursive calls.
841
def add_ancestors(file_id):
842
if file_id not in byid:
844
parent_id = byid[file_id].parent_id
845
if parent_id is None:
847
if parent_id not in parents:
848
parents.add(parent_id)
849
add_ancestors(parent_id)
850
for file_id in specific_file_ids:
851
add_ancestors(file_id)
855
stack = [(u'', from_dir)]
857
cur_relpath, cur_dir = stack.pop()
860
for child_name, child_ie in sorted(cur_dir.children.iteritems()):
862
child_relpath = cur_relpath + child_name
864
if (specific_file_ids is None or
865
child_ie.file_id in specific_file_ids or
866
(yield_parents and child_ie.file_id in parents)):
867
yield child_relpath, child_ie
869
if child_ie.kind == 'directory':
870
if parents is None or child_ie.file_id in parents:
871
child_dirs.append((child_relpath+'/', child_ie))
872
stack.extend(reversed(child_dirs))
874
def _make_delta(self, old):
875
"""Make an inventory delta from two inventories."""
878
adds = new_ids - old_ids
879
deletes = old_ids - new_ids
880
common = old_ids.intersection(new_ids)
882
for file_id in deletes:
883
delta.append((old.id2path(file_id), None, file_id, None))
885
delta.append((None, self.id2path(file_id), file_id, self[file_id]))
886
for file_id in common:
887
if old[file_id] != self[file_id]:
888
delta.append((old.id2path(file_id), self.id2path(file_id),
889
file_id, self[file_id]))
892
def _get_mutable_inventory(self):
893
"""Returns a mutable copy of the object.
895
Some inventories are immutable, yet working trees, for example, needs
896
to mutate exisiting inventories instead of creating a new one.
898
raise NotImplementedError(self._get_mutable_inventory)
900
def make_entry(self, kind, name, parent_id, file_id=None):
901
"""Simple thunk to bzrlib.inventory.make_entry."""
902
return make_entry(kind, name, parent_id, file_id)
905
"""Return list of (path, ie) for all entries except the root.
907
This may be faster than iter_entries.
910
def descend(dir_ie, dir_path):
911
kids = dir_ie.children.items()
913
for name, ie in kids:
914
child_path = osutils.pathjoin(dir_path, name)
915
accum.append((child_path, ie))
916
if ie.kind == 'directory':
917
descend(ie, child_path)
919
descend(self.root, u'')
922
def directories(self):
923
"""Return (path, entry) pairs for all directories, including the root.
926
def descend(parent_ie, parent_path):
927
accum.append((parent_path, parent_ie))
929
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
932
for name, child_ie in kids:
933
child_path = osutils.pathjoin(parent_path, name)
934
descend(child_ie, child_path)
935
descend(self.root, u'')
938
def path2id(self, name):
939
"""Walk down through directories to return entry of last component.
941
names may be either a list of path components, or a single
942
string, in which case it is automatically split.
944
This returns the entry of the last component in the path,
945
which may be either a file or a directory.
947
Returns None IFF the path is not found.
949
if isinstance(name, basestring):
950
name = osutils.splitpath(name)
952
# mutter("lookup path %r" % name)
956
except errors.NoSuchId:
957
# root doesn't exist yet so nothing else can
963
children = getattr(parent, 'children', None)
972
return parent.file_id
974
def filter(self, specific_fileids):
975
"""Get an inventory view filtered against a set of file-ids.
977
Children of directories and parents are included.
979
The result may or may not reference the underlying inventory
980
so it should be treated as immutable.
982
interesting_parents = set()
983
for fileid in specific_fileids:
985
interesting_parents.update(self.get_idpath(fileid))
986
except errors.NoSuchId:
987
# This fileid is not in the inventory - that's ok
989
entries = self.iter_entries()
990
if self.root is None:
991
return Inventory(root_id=None)
992
other = Inventory(entries.next()[1].file_id)
993
other.root.revision = self.root.revision
994
other.revision_id = self.revision_id
995
directories_to_expand = set()
996
for path, entry in entries:
997
file_id = entry.file_id
998
if (file_id in specific_fileids
999
or entry.parent_id in directories_to_expand):
1000
if entry.kind == 'directory':
1001
directories_to_expand.add(file_id)
1002
elif file_id not in interesting_parents:
1004
other.add(entry.copy())
1007
def get_idpath(self, file_id):
1008
"""Return a list of file_ids for the path to an entry.
1010
The list contains one element for each directory followed by
1011
the id of the file itself. So the length of the returned list
1012
is equal to the depth of the file in the tree, counting the
1013
root directory as depth 1.
1016
for parent in self._iter_file_id_parents(file_id):
1017
p.insert(0, parent.file_id)
1021
class Inventory(CommonInventory):
1022
"""Inventory of versioned files in a tree.
1024
This describes which file_id is present at each point in the tree,
1025
and possibly the SHA-1 or other information about the file.
1026
Entries can be looked up either by path or by file_id.
1028
The inventory represents a typical unix file tree, with
1029
directories containing files and subdirectories. We never store
1030
the full path to a file, because renaming a directory implicitly
1031
moves all of its contents. This class internally maintains a
1032
lookup tree that allows the children under a directory to be
1035
InventoryEntry objects must not be modified after they are
1036
inserted, other than through the Inventory API.
1038
>>> inv = Inventory()
1039
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
1040
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
1041
>>> inv['123-123'].name
1044
May be treated as an iterator or set to look up file ids:
1046
>>> bool(inv.path2id('hello.c'))
1048
>>> '123-123' in inv
1051
May also look up by name:
1053
>>> [x[0] for x in inv.iter_entries()]
1055
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
1056
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
1057
Traceback (most recent call last):
1058
BzrError: parent_id {TREE_ROOT} not in inventory
1059
>>> inv.add(InventoryFile('123-123', 'hello.c', 'TREE_ROOT-12345678-12345678'))
1060
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT-12345678-12345678', sha1=None, len=None, revision=None)
1062
def __init__(self, root_id=ROOT_ID, revision_id=None):
1063
"""Create or read an inventory.
1065
If a working directory is specified, the inventory is read
1066
from there. If the file is specified, read from that. If not,
1067
the inventory is created empty.
1069
The inventory is created with a default root directory, with
1072
if root_id is not None:
1073
self._set_root(InventoryDirectory(root_id, u'', None))
1077
self.revision_id = revision_id
1080
# More than one page of ouput is not useful anymore to debug
1083
contents = repr(self._byid)
1084
if len(contents) > max_len:
1085
contents = contents[:(max_len-len(closing))] + closing
1086
return "<Inventory object at %x, contents=%r>" % (id(self), contents)
1088
def apply_delta(self, delta):
1089
"""Apply a delta to this inventory.
1091
:param delta: A list of changes to apply. After all the changes are
1092
applied the final inventory must be internally consistent, but it
1093
is ok to supply changes which, if only half-applied would have an
1094
invalid result - such as supplying two changes which rename two
1095
files, 'A' and 'B' with each other : [('A', 'B', 'A-id', a_entry),
1096
('B', 'A', 'B-id', b_entry)].
1098
Each change is a tuple, of the form (old_path, new_path, file_id,
1101
When new_path is None, the change indicates the removal of an entry
1102
from the inventory and new_entry will be ignored (using None is
1103
appropriate). If new_path is not None, then new_entry must be an
1104
InventoryEntry instance, which will be incorporated into the
1105
inventory (and replace any existing entry with the same file id).
1107
When old_path is None, the change indicates the addition of
1108
a new entry to the inventory.
1110
When neither new_path nor old_path are None, the change is a
1111
modification to an entry, such as a rename, reparent, kind change
1114
The children attribute of new_entry is ignored. This is because
1115
this method preserves children automatically across alterations to
1116
the parent of the children, and cases where the parent id of a
1117
child is changing require the child to be passed in as a separate
1118
change regardless. E.g. in the recursive deletion of a directory -
1119
the directory's children must be included in the delta, or the
1120
final inventory will be invalid.
1122
Note that a file_id must only appear once within a given delta.
1123
An AssertionError is raised otherwise.
1125
# Check that the delta is legal. It would be nice if this could be
1126
# done within the loops below but it's safer to validate the delta
1127
# before starting to mutate the inventory.
1128
unique_file_ids = set([f for _, _, f, _ in delta])
1129
if len(unique_file_ids) != len(delta):
1130
raise AssertionError("a file-id appears multiple times in %r"
1135
# Remove all affected items which were in the original inventory,
1136
# starting with the longest paths, thus ensuring parents are examined
1137
# after their children, which means that everything we examine has no
1138
# modified children remaining by the time we examine it.
1139
for old_path, file_id in sorted(((op, f) for op, np, f, e in delta
1140
if op is not None), reverse=True):
1141
if file_id not in self:
1144
# Preserve unaltered children of file_id for later reinsertion.
1145
file_id_children = getattr(self[file_id], 'children', {})
1146
if len(file_id_children):
1147
children[file_id] = file_id_children
1148
# Remove file_id and the unaltered children. If file_id is not
1149
# being deleted it will be reinserted back later.
1150
self.remove_recursive_id(file_id)
1151
# Insert all affected which should be in the new inventory, reattaching
1152
# their children if they had any. This is done from shortest path to
1153
# longest, ensuring that items which were modified and whose parents in
1154
# the resulting inventory were also modified, are inserted after their
1156
for new_path, new_entry in sorted((np, e) for op, np, f, e in
1157
delta if np is not None):
1158
if new_entry.kind == 'directory':
1159
# Pop the child which to allow detection of children whose
1160
# parents were deleted and which were not reattached to a new
1162
replacement = InventoryDirectory(new_entry.file_id,
1163
new_entry.name, new_entry.parent_id)
1164
replacement.revision = new_entry.revision
1165
replacement.children = children.pop(replacement.file_id, {})
1166
new_entry = replacement
1169
# Get the parent id that was deleted
1170
parent_id, children = children.popitem()
1171
raise errors.InconsistentDelta("<deleted>", parent_id,
1172
"The file id was deleted but its children were not deleted.")
1174
def _set_root(self, ie):
1176
self._byid = {self.root.file_id: self.root}
1179
# TODO: jam 20051218 Should copy also copy the revision_id?
1180
entries = self.iter_entries()
1181
if self.root is None:
1182
return Inventory(root_id=None)
1183
other = Inventory(entries.next()[1].file_id)
1184
other.root.revision = self.root.revision
1185
# copy recursively so we know directories will be added before
1186
# their children. There are more efficient ways than this...
1187
for path, entry in entries:
1188
other.add(entry.copy())
1191
def _get_mutable_inventory(self):
1192
"""See CommonInventory._get_mutable_inventory."""
1193
return copy.deepcopy(self)
1196
"""Iterate over all file-ids."""
1197
return iter(self._byid)
1199
def iter_just_entries(self):
1200
"""Iterate over all entries.
1202
Unlike iter_entries(), just the entries are returned (not (path, ie))
1203
and the order of entries is undefined.
1205
XXX: We may not want to merge this into bzr.dev.
1207
if self.root is None:
1209
for _, ie in self._byid.iteritems():
1213
"""Returns number of entries."""
1214
return len(self._byid)
1216
def __getitem__(self, file_id):
1217
"""Return the entry for given file_id.
1219
>>> inv = Inventory()
1220
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
1221
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
1222
>>> inv['123123'].name
1226
return self._byid[file_id]
1228
# really we're passing an inventory, not a tree...
1229
raise errors.NoSuchId(self, file_id)
1231
def get_file_kind(self, file_id):
1232
return self._byid[file_id].kind
1234
def get_child(self, parent_id, filename):
1235
return self[parent_id].children.get(filename)
1237
def _add_child(self, entry):
1238
"""Add an entry to the inventory, without adding it to its parent"""
1239
if entry.file_id in self._byid:
1240
raise BzrError("inventory already contains entry with id {%s}" %
1242
self._byid[entry.file_id] = entry
1243
for child in getattr(entry, 'children', {}).itervalues():
1244
self._add_child(child)
1247
def add(self, entry):
1248
"""Add entry to inventory.
1250
To add a file to a branch ready to be committed, use Branch.add,
1253
Returns the new entry object.
1255
if entry.file_id in self._byid:
1256
raise errors.DuplicateFileId(entry.file_id,
1257
self._byid[entry.file_id])
1259
if entry.parent_id is None:
1263
parent = self._byid[entry.parent_id]
1265
raise BzrError("parent_id {%s} not in inventory" %
1268
if entry.name in parent.children:
1269
raise BzrError("%s is already versioned" %
1270
osutils.pathjoin(self.id2path(parent.file_id),
1271
entry.name).encode('utf-8'))
1272
parent.children[entry.name] = entry
1273
return self._add_child(entry)
1275
def add_path(self, relpath, kind, file_id=None, parent_id=None):
1276
"""Add entry from a path.
1278
The immediate parent must already be versioned.
1280
Returns the new entry object."""
1282
parts = osutils.splitpath(relpath)
1286
file_id = generate_ids.gen_root_id()
1287
self.root = InventoryDirectory(file_id, '', None)
1288
self._byid = {self.root.file_id: self.root}
1291
parent_path = parts[:-1]
1292
parent_id = self.path2id(parent_path)
1293
if parent_id is None:
1294
raise errors.NotVersionedError(path=parent_path)
1295
ie = make_entry(kind, parts[-1], parent_id, file_id)
1298
def __delitem__(self, file_id):
1299
"""Remove entry by id.
1301
>>> inv = Inventory()
1302
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1303
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
1311
del self._byid[file_id]
1312
if ie.parent_id is not None:
1313
del self[ie.parent_id].children[ie.name]
1315
def __eq__(self, other):
1316
"""Compare two sets by comparing their contents.
1318
>>> i1 = Inventory()
1319
>>> i2 = Inventory()
1322
>>> i1.add(InventoryFile('123', 'foo', ROOT_ID))
1323
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
1326
>>> i2.add(InventoryFile('123', 'foo', ROOT_ID))
1327
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
1331
if not isinstance(other, Inventory):
1332
return NotImplemented
1334
return self._byid == other._byid
1336
def __ne__(self, other):
1337
return not self.__eq__(other)
1340
raise ValueError('not hashable')
1342
def _iter_file_id_parents(self, file_id):
1343
"""Yield the parents of file_id up to the root."""
1344
while file_id is not None:
1346
ie = self._byid[file_id]
1348
raise errors.NoSuchId(tree=None, file_id=file_id)
1350
file_id = ie.parent_id
1352
def has_filename(self, filename):
1353
return bool(self.path2id(filename))
1355
def has_id(self, file_id):
1356
return (file_id in self._byid)
1358
def _make_delta(self, old):
1359
"""Make an inventory delta from two inventories."""
1360
old_getter = getattr(old, '_byid', old)
1361
new_getter = self._byid
1362
old_ids = set(old_getter)
1363
new_ids = set(new_getter)
1364
adds = new_ids - old_ids
1365
deletes = old_ids - new_ids
1366
if not adds and not deletes:
1369
common = old_ids.intersection(new_ids)
1371
for file_id in deletes:
1372
delta.append((old.id2path(file_id), None, file_id, None))
1373
for file_id in adds:
1374
delta.append((None, self.id2path(file_id), file_id, self[file_id]))
1375
for file_id in common:
1376
new_ie = new_getter[file_id]
1377
old_ie = old_getter[file_id]
1378
# If xml_serializer returns the cached InventoryEntries (rather
1379
# than always doing .copy()), inlining the 'is' check saves 2.7M
1380
# calls to __eq__. Under lsprof this saves 20s => 6s.
1381
# It is a minor improvement without lsprof.
1382
if old_ie is new_ie or old_ie == new_ie:
1385
delta.append((old.id2path(file_id), self.id2path(file_id),
1389
def remove_recursive_id(self, file_id):
1390
"""Remove file_id, and children, from the inventory.
1392
:param file_id: A file_id to remove.
1394
to_find_delete = [self._byid[file_id]]
1396
while to_find_delete:
1397
ie = to_find_delete.pop()
1398
to_delete.append(ie.file_id)
1399
if ie.kind == 'directory':
1400
to_find_delete.extend(ie.children.values())
1401
for file_id in reversed(to_delete):
1403
del self._byid[file_id]
1404
if ie.parent_id is not None:
1405
del self[ie.parent_id].children[ie.name]
1409
def rename(self, file_id, new_parent_id, new_name):
1410
"""Move a file within the inventory.
1412
This can change either the name, or the parent, or both.
1414
This does not move the working file.
1416
new_name = ensure_normalized_name(new_name)
1417
if not is_valid_name(new_name):
1418
raise BzrError("not an acceptable filename: %r" % new_name)
1420
new_parent = self._byid[new_parent_id]
1421
if new_name in new_parent.children:
1422
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
1424
new_parent_idpath = self.get_idpath(new_parent_id)
1425
if file_id in new_parent_idpath:
1426
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
1427
% (self.id2path(file_id), self.id2path(new_parent_id)))
1429
file_ie = self._byid[file_id]
1430
old_parent = self._byid[file_ie.parent_id]
1432
# TODO: Don't leave things messed up if this fails
1434
del old_parent.children[file_ie.name]
1435
new_parent.children[new_name] = file_ie
1437
file_ie.name = new_name
1438
file_ie.parent_id = new_parent_id
1440
def is_root(self, file_id):
1441
return self.root is not None and file_id == self.root.file_id
1444
class CHKInventory(CommonInventory):
1445
"""An inventory persisted in a CHK store.
1447
By design, a CHKInventory is immutable so many of the methods
1448
supported by Inventory - add, rename, apply_delta, etc - are *not*
1449
supported. To create a new CHKInventory, use create_by_apply_delta()
1450
or from_inventory(), say.
1452
Internally, a CHKInventory has one or two CHKMaps:
1454
* id_to_entry - a map from (file_id,) => InventoryEntry as bytes
1455
* parent_id_basename_to_file_id - a map from (parent_id, basename_utf8)
1458
The second map is optional and not present in early CHkRepository's.
1460
No caching is performed: every method call or item access will perform
1461
requests to the storage layer. As such, keep references to objects you
1465
def __init__(self, search_key_name):
1466
CommonInventory.__init__(self)
1467
self._fileid_to_entry_cache = {}
1468
self._path_to_fileid_cache = {}
1469
self._search_key_name = search_key_name
1471
def __eq__(self, other):
1472
"""Compare two sets by comparing their contents."""
1473
if not isinstance(other, CHKInventory):
1474
return NotImplemented
1476
this_key = self.id_to_entry.key()
1477
other_key = other.id_to_entry.key()
1478
this_pid_key = self.parent_id_basename_to_file_id.key()
1479
other_pid_key = other.parent_id_basename_to_file_id.key()
1480
if None in (this_key, this_pid_key, other_key, other_pid_key):
1482
return this_key == other_key and this_pid_key == other_pid_key
1484
def _entry_to_bytes(self, entry):
1485
"""Serialise entry as a single bytestring.
1487
:param Entry: An inventory entry.
1488
:return: A bytestring for the entry.
1491
ENTRY ::= FILE | DIR | SYMLINK | TREE
1492
FILE ::= "file: " COMMON SEP SHA SEP SIZE SEP EXECUTABLE
1493
DIR ::= "dir: " COMMON
1494
SYMLINK ::= "symlink: " COMMON SEP TARGET_UTF8
1495
TREE ::= "tree: " COMMON REFERENCE_REVISION
1496
COMMON ::= FILE_ID SEP PARENT_ID SEP NAME_UTF8 SEP REVISION
1499
if entry.parent_id is not None:
1500
parent_str = entry.parent_id
1503
name_str = entry.name.encode("utf8")
1504
if entry.kind == 'file':
1505
if entry.executable:
1509
return "file: %s\n%s\n%s\n%s\n%s\n%d\n%s" % (
1510
entry.file_id, parent_str, name_str, entry.revision,
1511
entry.text_sha1, entry.text_size, exec_str)
1512
elif entry.kind == 'directory':
1513
return "dir: %s\n%s\n%s\n%s" % (
1514
entry.file_id, parent_str, name_str, entry.revision)
1515
elif entry.kind == 'symlink':
1516
return "symlink: %s\n%s\n%s\n%s\n%s" % (
1517
entry.file_id, parent_str, name_str, entry.revision,
1518
entry.symlink_target.encode("utf8"))
1519
elif entry.kind == 'tree-reference':
1520
return "tree: %s\n%s\n%s\n%s\n%s" % (
1521
entry.file_id, parent_str, name_str, entry.revision,
1522
entry.reference_revision)
1524
raise ValueError("unknown kind %r" % entry.kind)
1527
def _bytes_to_utf8name_key(bytes):
1528
"""Get the file_id, revision_id key out of bytes."""
1529
# We don't normally care about name, except for times when we want
1530
# to filter out empty names because of non rich-root...
1531
sections = bytes.split('\n')
1532
kind, file_id = sections[0].split(': ')
1533
return (sections[2], file_id, sections[3])
1535
def _bytes_to_entry(self, bytes):
1536
"""Deserialise a serialised entry."""
1537
sections = bytes.split('\n')
1538
if sections[0].startswith("file: "):
1539
result = InventoryFile(sections[0][6:],
1540
sections[2].decode('utf8'),
1542
result.text_sha1 = sections[4]
1543
result.text_size = int(sections[5])
1544
result.executable = sections[6] == "Y"
1545
elif sections[0].startswith("dir: "):
1546
result = CHKInventoryDirectory(sections[0][5:],
1547
sections[2].decode('utf8'),
1549
elif sections[0].startswith("symlink: "):
1550
result = InventoryLink(sections[0][9:],
1551
sections[2].decode('utf8'),
1553
result.symlink_target = sections[4].decode('utf8')
1554
elif sections[0].startswith("tree: "):
1555
result = TreeReference(sections[0][6:],
1556
sections[2].decode('utf8'),
1558
result.reference_revision = sections[4]
1560
raise ValueError("Not a serialised entry %r" % bytes)
1561
result.revision = sections[3]
1562
if result.parent_id == '':
1563
result.parent_id = None
1564
self._fileid_to_entry_cache[result.file_id] = result
1567
def _get_mutable_inventory(self):
1568
"""See CommonInventory._get_mutable_inventory."""
1569
entries = self.iter_entries()
1570
inv = Inventory(None, self.revision_id)
1571
for path, inv_entry in entries:
1572
inv.add(inv_entry.copy())
1575
def create_by_apply_delta(self, inventory_delta, new_revision_id,
1576
propagate_caches=False):
1577
"""Create a new CHKInventory by applying inventory_delta to this one.
1579
:param inventory_delta: The inventory delta to apply. See
1580
Inventory.apply_delta for details.
1581
:param new_revision_id: The revision id of the resulting CHKInventory.
1582
:param propagate_caches: If True, the caches for this inventory are
1583
copied to and updated for the result.
1584
:return: The new CHKInventory.
1586
result = CHKInventory(self._search_key_name)
1587
if propagate_caches:
1588
# Just propagate the path-to-fileid cache for now
1589
result._path_to_fileid_cache = dict(self._path_to_fileid_cache.iteritems())
1590
search_key_func = chk_map.search_key_registry.get(self._search_key_name)
1591
self.id_to_entry._ensure_root()
1592
maximum_size = self.id_to_entry._root_node.maximum_size
1593
result.revision_id = new_revision_id
1594
result.id_to_entry = chk_map.CHKMap(
1595
self.id_to_entry._store,
1596
self.id_to_entry.key(),
1597
search_key_func=search_key_func)
1598
result.id_to_entry._ensure_root()
1599
result.id_to_entry._root_node.set_maximum_size(maximum_size)
1600
parent_id_basename_delta = []
1601
if self.parent_id_basename_to_file_id is not None:
1602
result.parent_id_basename_to_file_id = chk_map.CHKMap(
1603
self.parent_id_basename_to_file_id._store,
1604
self.parent_id_basename_to_file_id.key(),
1605
search_key_func=search_key_func)
1606
result.parent_id_basename_to_file_id._ensure_root()
1607
self.parent_id_basename_to_file_id._ensure_root()
1608
result_p_id_root = result.parent_id_basename_to_file_id._root_node
1609
p_id_root = self.parent_id_basename_to_file_id._root_node
1610
result_p_id_root.set_maximum_size(p_id_root.maximum_size)
1611
result_p_id_root._key_width = p_id_root._key_width
1613
result.parent_id_basename_to_file_id = None
1614
result.root_id = self.root_id
1615
id_to_entry_delta = []
1616
for old_path, new_path, file_id, entry in inventory_delta:
1619
result.root_id = file_id
1620
if new_path is None:
1625
if propagate_caches:
1627
del result._path_to_fileid_cache[old_path]
1631
new_key = (file_id,)
1632
new_value = result._entry_to_bytes(entry)
1633
# Update caches. It's worth doing this whether
1634
# we're propagating the old caches or not.
1635
result._path_to_fileid_cache[new_path] = file_id
1636
if old_path is None:
1639
old_key = (file_id,)
1640
id_to_entry_delta.append((old_key, new_key, new_value))
1641
if result.parent_id_basename_to_file_id is not None:
1642
# parent_id, basename changes
1643
if old_path is None:
1646
old_entry = self[file_id]
1647
old_key = self._parent_id_basename_key(old_entry)
1648
if new_path is None:
1652
new_key = self._parent_id_basename_key(entry)
1654
if old_key != new_key:
1655
# If the two keys are the same, the value will be unchanged
1656
# as its always the file id.
1657
parent_id_basename_delta.append((old_key, new_key, new_value))
1658
result.id_to_entry.apply_delta(id_to_entry_delta)
1659
if parent_id_basename_delta:
1660
result.parent_id_basename_to_file_id.apply_delta(parent_id_basename_delta)
1664
def deserialise(klass, chk_store, bytes, expected_revision_id):
1665
"""Deserialise a CHKInventory.
1667
:param chk_store: A CHK capable VersionedFiles instance.
1668
:param bytes: The serialised bytes.
1669
:param expected_revision_id: The revision ID we think this inventory is
1671
:return: A CHKInventory
1673
lines = bytes.split('\n')
1675
raise AssertionError('bytes to deserialize must end with an eol')
1677
if lines[0] != 'chkinventory:':
1678
raise ValueError("not a serialised CHKInventory: %r" % bytes)
1680
allowed_keys = frozenset(['root_id', 'revision_id', 'search_key_name',
1681
'parent_id_basename_to_file_id',
1683
for line in lines[1:]:
1684
key, value = line.split(': ', 1)
1685
if key not in allowed_keys:
1686
raise errors.BzrError('Unknown key in inventory: %r\n%r'
1689
raise errors.BzrError('Duplicate key in inventory: %r\n%r'
1692
revision_id = info['revision_id']
1693
root_id = info['root_id']
1694
search_key_name = info.get('search_key_name', 'plain')
1695
parent_id_basename_to_file_id = info.get(
1696
'parent_id_basename_to_file_id', None)
1697
id_to_entry = info['id_to_entry']
1699
result = CHKInventory(search_key_name)
1700
result.revision_id = revision_id
1701
result.root_id = root_id
1702
search_key_func = chk_map.search_key_registry.get(
1703
result._search_key_name)
1704
if parent_id_basename_to_file_id is not None:
1705
result.parent_id_basename_to_file_id = chk_map.CHKMap(
1706
chk_store, (parent_id_basename_to_file_id,),
1707
search_key_func=search_key_func)
1709
result.parent_id_basename_to_file_id = None
1711
result.id_to_entry = chk_map.CHKMap(chk_store, (id_to_entry,),
1712
search_key_func=search_key_func)
1713
if (result.revision_id,) != expected_revision_id:
1714
raise ValueError("Mismatched revision id and expected: %r, %r" %
1715
(result.revision_id, expected_revision_id))
1719
def from_inventory(klass, chk_store, inventory, maximum_size=0, search_key_name='plain'):
1720
"""Create a CHKInventory from an existing inventory.
1722
The content of inventory is copied into the chk_store, and a
1723
CHKInventory referencing that is returned.
1725
:param chk_store: A CHK capable VersionedFiles instance.
1726
:param inventory: The inventory to copy.
1727
:param maximum_size: The CHKMap node size limit.
1728
:param search_key_name: The identifier for the search key function
1730
result = klass(search_key_name)
1731
result.revision_id = inventory.revision_id
1732
result.root_id = inventory.root.file_id
1734
entry_to_bytes = result._entry_to_bytes
1735
parent_id_basename_key = result._parent_id_basename_key
1736
id_to_entry_dict = {}
1737
parent_id_basename_dict = {}
1738
for path, entry in inventory.iter_entries():
1739
id_to_entry_dict[(entry.file_id,)] = entry_to_bytes(entry)
1740
p_id_key = parent_id_basename_key(entry)
1741
parent_id_basename_dict[p_id_key] = entry.file_id
1743
result._populate_from_dicts(chk_store, id_to_entry_dict,
1744
parent_id_basename_dict, maximum_size=maximum_size)
1747
def _populate_from_dicts(self, chk_store, id_to_entry_dict,
1748
parent_id_basename_dict, maximum_size):
1749
search_key_func = chk_map.search_key_registry.get(self._search_key_name)
1750
root_key = chk_map.CHKMap.from_dict(chk_store, id_to_entry_dict,
1751
maximum_size=maximum_size, key_width=1,
1752
search_key_func=search_key_func)
1753
self.id_to_entry = chk_map.CHKMap(chk_store, root_key,
1755
root_key = chk_map.CHKMap.from_dict(chk_store,
1756
parent_id_basename_dict,
1757
maximum_size=maximum_size, key_width=2,
1758
search_key_func=search_key_func)
1759
self.parent_id_basename_to_file_id = chk_map.CHKMap(chk_store,
1760
root_key, search_key_func)
1762
def _parent_id_basename_key(self, entry):
1763
"""Create a key for a entry in a parent_id_basename_to_file_id index."""
1764
if entry.parent_id is not None:
1765
parent_id = entry.parent_id
1768
return parent_id, entry.name.encode('utf8')
1770
def __getitem__(self, file_id):
1771
"""map a single file_id -> InventoryEntry."""
1773
raise errors.NoSuchId(self, file_id)
1774
result = self._fileid_to_entry_cache.get(file_id, None)
1775
if result is not None:
1778
return self._bytes_to_entry(
1779
self.id_to_entry.iteritems([(file_id,)]).next()[1])
1780
except StopIteration:
1781
# really we're passing an inventory, not a tree...
1782
raise errors.NoSuchId(self, file_id)
1784
def has_id(self, file_id):
1785
# Perhaps have an explicit 'contains' method on CHKMap ?
1786
if self._fileid_to_entry_cache.get(file_id, None) is not None:
1788
return len(list(self.id_to_entry.iteritems([(file_id,)]))) == 1
1790
def is_root(self, file_id):
1791
return file_id == self.root_id
1793
def _iter_file_id_parents(self, file_id):
1794
"""Yield the parents of file_id up to the root."""
1795
while file_id is not None:
1799
raise errors.NoSuchId(tree=self, file_id=file_id)
1801
file_id = ie.parent_id
1804
"""Iterate over all file-ids."""
1805
for key, _ in self.id_to_entry.iteritems():
1808
def iter_just_entries(self):
1809
"""Iterate over all entries.
1811
Unlike iter_entries(), just the entries are returned (not (path, ie))
1812
and the order of entries is undefined.
1814
XXX: We may not want to merge this into bzr.dev.
1816
for key, entry in self.id_to_entry.iteritems():
1818
ie = self._fileid_to_entry_cache.get(file_id, None)
1820
ie = self._bytes_to_entry(entry)
1821
self._fileid_to_entry_cache[file_id] = ie
1824
def iter_changes(self, basis):
1825
"""Generate a Tree.iter_changes change list between this and basis.
1827
:param basis: Another CHKInventory.
1828
:return: An iterator over the changes between self and basis, as per
1829
tree.iter_changes().
1831
# We want: (file_id, (path_in_source, path_in_target),
1832
# changed_content, versioned, parent, name, kind,
1834
for key, basis_value, self_value in \
1835
self.id_to_entry.iter_changes(basis.id_to_entry):
1837
if basis_value is not None:
1838
basis_entry = basis._bytes_to_entry(basis_value)
1839
path_in_source = basis.id2path(file_id)
1840
basis_parent = basis_entry.parent_id
1841
basis_name = basis_entry.name
1842
basis_executable = basis_entry.executable
1844
path_in_source = None
1847
basis_executable = None
1848
if self_value is not None:
1849
self_entry = self._bytes_to_entry(self_value)
1850
path_in_target = self.id2path(file_id)
1851
self_parent = self_entry.parent_id
1852
self_name = self_entry.name
1853
self_executable = self_entry.executable
1855
path_in_target = None
1858
self_executable = None
1859
if basis_value is None:
1861
kind = (None, self_entry.kind)
1862
versioned = (False, True)
1863
elif self_value is None:
1865
kind = (basis_entry.kind, None)
1866
versioned = (True, False)
1868
kind = (basis_entry.kind, self_entry.kind)
1869
versioned = (True, True)
1870
changed_content = False
1871
if kind[0] != kind[1]:
1872
changed_content = True
1873
elif kind[0] == 'file':
1874
if (self_entry.text_size != basis_entry.text_size or
1875
self_entry.text_sha1 != basis_entry.text_sha1):
1876
changed_content = True
1877
elif kind[0] == 'symlink':
1878
if self_entry.symlink_target != basis_entry.symlink_target:
1879
changed_content = True
1880
elif kind[0] == 'tree-reference':
1881
if (self_entry.reference_revision !=
1882
basis_entry.reference_revision):
1883
changed_content = True
1884
parent = (basis_parent, self_parent)
1885
name = (basis_name, self_name)
1886
executable = (basis_executable, self_executable)
1887
if (not changed_content
1888
and parent[0] == parent[1]
1889
and name[0] == name[1]
1890
and executable[0] == executable[1]):
1891
# Could happen when only the revision changed for a directory
1894
yield (file_id, (path_in_source, path_in_target), changed_content,
1895
versioned, parent, name, kind, executable)
1898
"""Return the number of entries in the inventory."""
1899
return len(self.id_to_entry)
1901
def _make_delta(self, old):
1902
"""Make an inventory delta from two inventories."""
1903
if type(old) != CHKInventory:
1904
return CommonInventory._make_delta(self, old)
1906
for key, old_value, self_value in \
1907
self.id_to_entry.iter_changes(old.id_to_entry):
1909
if old_value is not None:
1910
old_path = old.id2path(file_id)
1913
if self_value is not None:
1914
entry = self._bytes_to_entry(self_value)
1915
self._fileid_to_entry_cache[file_id] = entry
1916
new_path = self.id2path(file_id)
1920
delta.append((old_path, new_path, file_id, entry))
1923
def path2id(self, name):
1924
"""See CommonInventory.path2id()."""
1925
result = self._path_to_fileid_cache.get(name, None)
1927
result = CommonInventory.path2id(self, name)
1928
self._path_to_fileid_cache[name] = result
1932
"""Serialise the inventory to lines."""
1933
lines = ["chkinventory:\n"]
1934
if self._search_key_name != 'plain':
1935
# custom ordering grouping things that don't change together
1936
lines.append('search_key_name: %s\n' % (self._search_key_name,))
1937
lines.append("root_id: %s\n" % self.root_id)
1938
lines.append('parent_id_basename_to_file_id: %s\n' %
1939
self.parent_id_basename_to_file_id.key())
1940
lines.append("revision_id: %s\n" % self.revision_id)
1941
lines.append("id_to_entry: %s\n" % self.id_to_entry.key())
1943
lines.append("revision_id: %s\n" % self.revision_id)
1944
lines.append("root_id: %s\n" % self.root_id)
1945
if self.parent_id_basename_to_file_id is not None:
1946
lines.append('parent_id_basename_to_file_id: %s\n' %
1947
self.parent_id_basename_to_file_id.key())
1948
lines.append("id_to_entry: %s\n" % self.id_to_entry.key())
1953
"""Get the root entry."""
1954
return self[self.root_id]
1957
class CHKInventoryDirectory(InventoryDirectory):
1958
"""A directory in an inventory."""
1960
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
1961
'text_id', 'parent_id', '_children', 'executable',
1962
'revision', 'symlink_target', 'reference_revision',
1965
def __init__(self, file_id, name, parent_id, chk_inventory):
1966
# Don't call InventoryDirectory.__init__ - it isn't right for this
1968
InventoryEntry.__init__(self, file_id, name, parent_id)
1969
self._children = None
1970
self.kind = 'directory'
1971
self._chk_inventory = chk_inventory
1975
"""Access the list of children of this directory.
1977
With a parent_id_basename_to_file_id index, loads all the children,
1978
without loads the entire index. Without is bad. A more sophisticated
1979
proxy object might be nice, to allow partial loading of children as
1980
well when specific names are accessed. (So path traversal can be
1981
written in the obvious way but not examine siblings.).
1983
if self._children is not None:
1984
return self._children
1985
# No longer supported
1986
if self._chk_inventory.parent_id_basename_to_file_id is None:
1987
raise AssertionError("Inventories without"
1988
" parent_id_basename_to_file_id are no longer supported")
1990
# XXX: Todo - use proxy objects for the children rather than loading
1991
# all when the attribute is referenced.
1992
parent_id_index = self._chk_inventory.parent_id_basename_to_file_id
1994
for (parent_id, name_utf8), file_id in parent_id_index.iteritems(
1995
key_filter=[(self.file_id,)]):
1996
child_keys.add((file_id,))
1998
for file_id_key in child_keys:
1999
entry = self._chk_inventory._fileid_to_entry_cache.get(
2000
file_id_key[0], None)
2001
if entry is not None:
2002
result[entry.name] = entry
2003
cached.add(file_id_key)
2004
child_keys.difference_update(cached)
2005
# populate; todo: do by name
2006
id_to_entry = self._chk_inventory.id_to_entry
2007
for file_id_key, bytes in id_to_entry.iteritems(child_keys):
2008
entry = self._chk_inventory._bytes_to_entry(bytes)
2009
result[entry.name] = entry
2010
self._chk_inventory._fileid_to_entry_cache[file_id_key[0]] = entry
2011
self._children = result
2015
'directory': InventoryDirectory,
2016
'file': InventoryFile,
2017
'symlink': InventoryLink,
2018
'tree-reference': TreeReference
2021
def make_entry(kind, name, parent_id, file_id=None):
2022
"""Create an inventory entry.
2024
:param kind: the type of inventory entry to create.
2025
:param name: the basename of the entry.
2026
:param parent_id: the parent_id of the entry.
2027
:param file_id: the file_id to use. if None, one will be created.
2030
file_id = generate_ids.gen_file_id(name)
2031
name = ensure_normalized_name(name)
2033
factory = entry_factory[kind]
2035
raise BzrError("unknown kind %r" % kind)
2036
return factory(file_id, name, parent_id)
2039
def ensure_normalized_name(name):
2042
:raises InvalidNormalization: When name is not normalized, and cannot be
2043
accessed on this platform by the normalized path.
2044
:return: The NFC normalised version of name.
2046
#------- This has been copied to bzrlib.dirstate.DirState.add, please
2047
# keep them synchronised.
2048
# we dont import normalized_filename directly because we want to be
2049
# able to change the implementation at runtime for tests.
2050
norm_name, can_access = osutils.normalized_filename(name)
2051
if norm_name != name:
2055
# TODO: jam 20060701 This would probably be more useful
2056
# if the error was raised with the full path
2057
raise errors.InvalidNormalization(name)
2063
def is_valid_name(name):
2065
if _NAME_RE is None:
2066
_NAME_RE = re.compile(r'^[^/\\]+$')
2068
return bool(_NAME_RE.match(name))
added
removed
bzrlib/inventory.py
