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 copy import deepcopy
32
from bzrlib.lazy_import import lazy_import
33
lazy_import(globals(), """
50
from bzrlib.errors import (
54
from bzrlib.symbol_versioning import deprecated_in, deprecated_method
55
from bzrlib.trace import mutter
58
class InventoryEntry(object):
59
"""Description of a versioned file.
61
An InventoryEntry has the following fields, which are also
62
present in the XML inventory-entry element:
67
(within the parent directory)
70
file_id of the parent directory, or ROOT_ID
73
the revision_id in which this variation of this file was
77
Indicates that this file should be executable on systems
81
sha-1 of the text of the file
84
size in bytes of the text of the file
86
(reading a version 4 tree created a text_id field.)
91
>>> i.add(InventoryDirectory('123', 'src', ROOT_ID))
92
InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None)
93
>>> i.add(InventoryFile('2323', 'hello.c', parent_id='123'))
94
InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None, revision=None)
95
>>> shouldbe = {0: '', 1: 'src', 2: 'src/hello.c'}
96
>>> for ix, j in enumerate(i.iter_entries()):
97
... print (j[0] == shouldbe[ix], j[1])
99
(True, InventoryDirectory('TREE_ROOT', u'', parent_id=None, revision=None))
100
(True, InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None))
101
(True, InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None, revision=None))
102
>>> i.add(InventoryFile('2324', 'bye.c', '123'))
103
InventoryFile('2324', 'bye.c', parent_id='123', sha1=None, len=None, revision=None)
104
>>> i.add(InventoryDirectory('2325', 'wibble', '123'))
105
InventoryDirectory('2325', 'wibble', parent_id='123', revision=None)
106
>>> i.path2id('src/wibble')
110
>>> i.add(InventoryFile('2326', 'wibble.c', '2325'))
111
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None, revision=None)
113
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None, revision=None)
114
>>> for path, entry in i.iter_entries():
123
>>> i.id2path('2326')
124
'src/wibble/wibble.c'
127
# Constants returned by describe_change()
129
# TODO: These should probably move to some kind of FileChangeDescription
130
# class; that's like what's inside a TreeDelta but we want to be able to
131
# generate them just for one file at a time.
133
MODIFIED_AND_RENAMED = 'modified and renamed'
137
def detect_changes(self, old_entry):
138
"""Return a (text_modified, meta_modified) from this to old_entry.
140
_read_tree_state must have been called on self and old_entry prior to
141
calling detect_changes.
145
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
146
output_to, reverse=False):
147
"""Perform a diff between two entries of the same kind."""
149
def parent_candidates(self, previous_inventories):
150
"""Find possible per-file graph parents.
152
This is currently defined by:
153
- Select the last changed revision in the parent inventory.
154
- Do deal with a short lived bug in bzr 0.8's development two entries
155
that have the same last changed but different 'x' bit settings are
158
# revision:ie mapping for each ie found in previous_inventories.
160
# identify candidate head revision ids.
161
for inv in previous_inventories:
162
if self.file_id in inv:
163
ie = inv[self.file_id]
164
if ie.revision in candidates:
165
# same revision value in two different inventories:
166
# correct possible inconsistencies:
167
# * there was a bug in revision updates with 'x' bit
170
if candidates[ie.revision].executable != ie.executable:
171
candidates[ie.revision].executable = False
172
ie.executable = False
173
except AttributeError:
176
# add this revision as a candidate.
177
candidates[ie.revision] = ie
180
@deprecated_method(deprecated_in((1, 6, 0)))
181
def get_tar_item(self, root, dp, now, tree):
182
"""Get a tarfile item and a file stream for its content."""
183
item = tarfile.TarInfo(osutils.pathjoin(root, dp).encode('utf8'))
184
# TODO: would be cool to actually set it to the timestamp of the
185
# revision it was last changed
187
fileobj = self._put_in_tar(item, tree)
191
"""Return true if the object this entry represents has textual data.
193
Note that textual data includes binary content.
195
Also note that all entries get weave files created for them.
196
This attribute is primarily used when upgrading from old trees that
197
did not have the weave index for all inventory entries.
201
def __init__(self, file_id, name, parent_id, text_id=None):
202
"""Create an InventoryEntry
204
The filename must be a single component, relative to the
205
parent directory; it cannot be a whole path or relative name.
207
>>> e = InventoryFile('123', 'hello.c', ROOT_ID)
212
>>> e = InventoryFile('123', 'src/hello.c', ROOT_ID)
213
Traceback (most recent call last):
214
InvalidEntryName: Invalid entry name: src/hello.c
216
if '/' in name or '\\' in name:
217
raise errors.InvalidEntryName(name=name)
218
self.executable = False
220
self.text_sha1 = None
221
self.text_size = None
222
self.file_id = file_id
224
self.text_id = text_id
225
self.parent_id = parent_id
226
self.symlink_target = None
227
self.reference_revision = None
229
def kind_character(self):
230
"""Return a short kind indicator useful for appending to names."""
231
raise BzrError('unknown kind %r' % self.kind)
233
known_kinds = ('file', 'directory', 'symlink')
235
def _put_in_tar(self, item, tree):
236
"""populate item for stashing in a tar, and return the content stream.
238
If no content is available, return None.
240
raise BzrError("don't know how to export {%s} of kind %r" %
241
(self.file_id, self.kind))
243
@deprecated_method(deprecated_in((1, 6, 0)))
244
def put_on_disk(self, dest, dp, tree):
245
"""Create a representation of self on disk in the prefix dest.
247
This is a template method - implement _put_on_disk in subclasses.
249
fullpath = osutils.pathjoin(dest, dp)
250
self._put_on_disk(fullpath, tree)
251
# mutter(" export {%s} kind %s to %s", self.file_id,
252
# self.kind, fullpath)
254
def _put_on_disk(self, fullpath, tree):
255
"""Put this entry onto disk at fullpath, from tree tree."""
256
raise BzrError("don't know how to export {%s} of kind %r" % (self.file_id, self.kind))
258
def sorted_children(self):
259
return sorted(self.children.items())
262
def versionable_kind(kind):
263
return (kind in ('file', 'directory', 'symlink', 'tree-reference'))
265
def check(self, checker, rev_id, inv, tree):
266
"""Check this inventory entry is intact.
268
This is a template method, override _check for kind specific
271
:param checker: Check object providing context for the checks;
272
can be used to find out what parts of the repository have already
274
:param rev_id: Revision id from which this InventoryEntry was loaded.
275
Not necessarily the last-changed revision for this file.
276
:param inv: Inventory from which the entry was loaded.
277
:param tree: RevisionTree for this entry.
279
if self.parent_id is not None:
280
if not inv.has_id(self.parent_id):
281
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
282
% (self.parent_id, rev_id))
283
self._check(checker, rev_id, tree)
285
def _check(self, checker, rev_id, tree):
286
"""Check this inventory entry for kind specific errors."""
287
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
291
"""Clone this inventory entry."""
292
raise NotImplementedError
295
def describe_change(old_entry, new_entry):
296
"""Describe the change between old_entry and this.
298
This smells of being an InterInventoryEntry situation, but as its
299
the first one, we're making it a static method for now.
301
An entry with a different parent, or different name is considered
302
to be renamed. Reparenting is an internal detail.
303
Note that renaming the parent does not trigger a rename for the
306
# TODO: Perhaps return an object rather than just a string
307
if old_entry is new_entry:
308
# also the case of both being None
310
elif old_entry is None:
312
elif new_entry is None:
314
if old_entry.kind != new_entry.kind:
316
text_modified, meta_modified = new_entry.detect_changes(old_entry)
317
if text_modified or meta_modified:
321
# TODO 20060511 (mbp, rbc) factor out 'detect_rename' here.
322
if old_entry.parent_id != new_entry.parent_id:
324
elif old_entry.name != new_entry.name:
328
if renamed and not modified:
329
return InventoryEntry.RENAMED
330
if modified and not renamed:
332
if modified and renamed:
333
return InventoryEntry.MODIFIED_AND_RENAMED
337
return ("%s(%r, %r, parent_id=%r, revision=%r)"
338
% (self.__class__.__name__,
344
def __eq__(self, other):
346
# For the case when objects are cached
348
if not isinstance(other, InventoryEntry):
349
return NotImplemented
351
return ((self.file_id == other.file_id)
352
and (self.name == other.name)
353
and (other.symlink_target == self.symlink_target)
354
and (self.text_sha1 == other.text_sha1)
355
and (self.text_size == other.text_size)
356
and (self.text_id == other.text_id)
357
and (self.parent_id == other.parent_id)
358
and (self.kind == other.kind)
359
and (self.revision == other.revision)
360
and (self.executable == other.executable)
361
and (self.reference_revision == other.reference_revision)
364
def __ne__(self, other):
365
return not (self == other)
368
raise ValueError('not hashable')
370
def _unchanged(self, previous_ie):
371
"""Has this entry changed relative to previous_ie.
373
This method should be overridden in child classes.
376
# different inv parent
377
if previous_ie.parent_id != self.parent_id:
380
elif previous_ie.name != self.name:
382
elif previous_ie.kind != self.kind:
386
def _read_tree_state(self, path, work_tree):
387
"""Populate fields in the inventory entry from the given tree.
389
Note that this should be modified to be a noop on virtual trees
390
as all entries created there are prepopulated.
392
# TODO: Rather than running this manually, we should check the
393
# working sha1 and other expensive properties when they're
394
# first requested, or preload them if they're already known
395
pass # nothing to do by default
397
def _forget_tree_state(self):
401
class RootEntry(InventoryEntry):
403
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
404
'text_id', 'parent_id', 'children', 'executable',
405
'revision', 'symlink_target', 'reference_revision']
407
def _check(self, checker, rev_id, tree):
408
"""See InventoryEntry._check"""
410
def __init__(self, file_id):
411
self.file_id = file_id
413
self.kind = 'directory'
414
self.parent_id = None
417
symbol_versioning.warn('RootEntry is deprecated as of bzr 0.10.'
418
' Please use InventoryDirectory instead.',
419
DeprecationWarning, stacklevel=2)
421
def __eq__(self, other):
422
if not isinstance(other, RootEntry):
423
return NotImplemented
425
return (self.file_id == other.file_id) \
426
and (self.children == other.children)
429
class InventoryDirectory(InventoryEntry):
430
"""A directory in an inventory."""
432
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
433
'text_id', 'parent_id', 'children', 'executable',
434
'revision', 'symlink_target', 'reference_revision']
436
def _check(self, checker, rev_id, tree):
437
"""See InventoryEntry._check"""
438
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
439
raise BzrCheckError('directory {%s} has text in revision {%s}'
440
% (self.file_id, rev_id))
443
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
444
other.revision = self.revision
445
# note that children are *not* copied; they're pulled across when
449
def __init__(self, file_id, name, parent_id):
450
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
452
self.kind = 'directory'
454
def kind_character(self):
455
"""See InventoryEntry.kind_character."""
458
def _put_in_tar(self, item, tree):
459
"""See InventoryEntry._put_in_tar."""
460
item.type = tarfile.DIRTYPE
467
def _put_on_disk(self, fullpath, tree):
468
"""See InventoryEntry._put_on_disk."""
472
class InventoryFile(InventoryEntry):
473
"""A file in an inventory."""
475
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
476
'text_id', 'parent_id', 'children', 'executable',
477
'revision', 'symlink_target', 'reference_revision']
479
def _check(self, checker, tree_revision_id, tree):
480
"""See InventoryEntry._check"""
481
key = (self.file_id, self.revision)
482
if key in checker.checked_texts:
483
prev_sha = checker.checked_texts[key]
484
if prev_sha != self.text_sha1:
486
'mismatched sha1 on {%s} in {%s} (%s != %s) %r' %
487
(self.file_id, tree_revision_id, prev_sha, self.text_sha1,
490
checker.repeated_text_cnt += 1
493
checker.checked_text_cnt += 1
494
# We can't check the length, because Weave doesn't store that
495
# information, and the whole point of looking at the weave's
496
# sha1sum is that we don't have to extract the text.
497
if (self.text_sha1 != tree._repository.texts.get_sha1s([key])[key]):
498
raise BzrCheckError('text {%s} version {%s} wrong sha1' % key)
499
checker.checked_texts[key] = self.text_sha1
502
other = InventoryFile(self.file_id, self.name, self.parent_id)
503
other.executable = self.executable
504
other.text_id = self.text_id
505
other.text_sha1 = self.text_sha1
506
other.text_size = self.text_size
507
other.revision = self.revision
510
def detect_changes(self, old_entry):
511
"""See InventoryEntry.detect_changes."""
512
text_modified = (self.text_sha1 != old_entry.text_sha1)
513
meta_modified = (self.executable != old_entry.executable)
514
return text_modified, meta_modified
516
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
517
output_to, reverse=False):
518
"""See InventoryEntry._diff."""
519
from bzrlib.diff import DiffText
520
from_file_id = self.file_id
522
to_file_id = to_entry.file_id
526
to_file_id, from_file_id = from_file_id, to_file_id
527
tree, to_tree = to_tree, tree
528
from_label, to_label = to_label, from_label
529
differ = DiffText(tree, to_tree, output_to, 'utf-8', '', '',
531
return differ.diff_text(from_file_id, to_file_id, from_label, to_label)
534
"""See InventoryEntry.has_text."""
537
def __init__(self, file_id, name, parent_id):
538
super(InventoryFile, self).__init__(file_id, name, parent_id)
541
def kind_character(self):
542
"""See InventoryEntry.kind_character."""
545
def _put_in_tar(self, item, tree):
546
"""See InventoryEntry._put_in_tar."""
547
item.type = tarfile.REGTYPE
548
fileobj = tree.get_file(self.file_id)
549
item.size = self.text_size
550
if tree.is_executable(self.file_id):
556
def _put_on_disk(self, fullpath, tree):
557
"""See InventoryEntry._put_on_disk."""
558
osutils.pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
559
if tree.is_executable(self.file_id):
560
os.chmod(fullpath, 0755)
562
def _read_tree_state(self, path, work_tree):
563
"""See InventoryEntry._read_tree_state."""
564
self.text_sha1 = work_tree.get_file_sha1(self.file_id, path=path)
565
# FIXME: 20050930 probe for the text size when getting sha1
566
# in _read_tree_state
567
self.executable = work_tree.is_executable(self.file_id, path=path)
570
return ("%s(%r, %r, parent_id=%r, sha1=%r, len=%s, revision=%s)"
571
% (self.__class__.__name__,
579
def _forget_tree_state(self):
580
self.text_sha1 = None
582
def _unchanged(self, previous_ie):
583
"""See InventoryEntry._unchanged."""
584
compatible = super(InventoryFile, self)._unchanged(previous_ie)
585
if self.text_sha1 != previous_ie.text_sha1:
588
# FIXME: 20050930 probe for the text size when getting sha1
589
# in _read_tree_state
590
self.text_size = previous_ie.text_size
591
if self.executable != previous_ie.executable:
596
class InventoryLink(InventoryEntry):
597
"""A file in an inventory."""
599
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
600
'text_id', 'parent_id', 'children', 'executable',
601
'revision', 'symlink_target', 'reference_revision']
603
def _check(self, checker, rev_id, tree):
604
"""See InventoryEntry._check"""
605
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
606
raise BzrCheckError('symlink {%s} has text in revision {%s}'
607
% (self.file_id, rev_id))
608
if self.symlink_target is None:
609
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
610
% (self.file_id, rev_id))
613
other = InventoryLink(self.file_id, self.name, self.parent_id)
614
other.symlink_target = self.symlink_target
615
other.revision = self.revision
618
def detect_changes(self, old_entry):
619
"""See InventoryEntry.detect_changes."""
620
# FIXME: which _modified field should we use ? RBC 20051003
621
text_modified = (self.symlink_target != old_entry.symlink_target)
623
mutter(" symlink target changed")
624
meta_modified = False
625
return text_modified, meta_modified
627
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
628
output_to, reverse=False):
629
"""See InventoryEntry._diff."""
630
from bzrlib.diff import DiffSymlink
631
old_target = self.symlink_target
632
if to_entry is not None:
633
new_target = to_entry.symlink_target
642
new_target, old_target = old_target, new_target
643
differ = DiffSymlink(old_tree, new_tree, output_to)
644
return differ.diff_symlink(old_target, new_target)
646
def __init__(self, file_id, name, parent_id):
647
super(InventoryLink, self).__init__(file_id, name, parent_id)
648
self.kind = 'symlink'
650
def kind_character(self):
651
"""See InventoryEntry.kind_character."""
654
def _put_in_tar(self, item, tree):
655
"""See InventoryEntry._put_in_tar."""
656
item.type = tarfile.SYMTYPE
660
item.linkname = self.symlink_target
663
def _put_on_disk(self, fullpath, tree):
664
"""See InventoryEntry._put_on_disk."""
666
os.symlink(self.symlink_target, fullpath)
668
raise BzrError("Failed to create symlink %r -> %r, error: %s" % (fullpath, self.symlink_target, e))
670
def _read_tree_state(self, path, work_tree):
671
"""See InventoryEntry._read_tree_state."""
672
self.symlink_target = work_tree.get_symlink_target(self.file_id)
674
def _forget_tree_state(self):
675
self.symlink_target = None
677
def _unchanged(self, previous_ie):
678
"""See InventoryEntry._unchanged."""
679
compatible = super(InventoryLink, self)._unchanged(previous_ie)
680
if self.symlink_target != previous_ie.symlink_target:
685
class TreeReference(InventoryEntry):
687
kind = 'tree-reference'
689
def __init__(self, file_id, name, parent_id, revision=None,
690
reference_revision=None):
691
InventoryEntry.__init__(self, file_id, name, parent_id)
692
self.revision = revision
693
self.reference_revision = reference_revision
696
return TreeReference(self.file_id, self.name, self.parent_id,
697
self.revision, self.reference_revision)
699
def _read_tree_state(self, path, work_tree):
700
"""Populate fields in the inventory entry from the given tree.
702
self.reference_revision = work_tree.get_reference_revision(
705
def _forget_tree_state(self):
706
self.reference_revision = None
708
def _unchanged(self, previous_ie):
709
"""See InventoryEntry._unchanged."""
710
compatible = super(TreeReference, self)._unchanged(previous_ie)
711
if self.reference_revision != previous_ie.reference_revision:
716
class CommonInventory(object):
717
"""Basic inventory logic, defined in terms of primitives like has_id."""
719
def __contains__(self, file_id):
720
"""True if this entry contains a file with given id.
722
>>> inv = Inventory()
723
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
724
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
730
Note that this method along with __iter__ are not encouraged for use as
731
they are less clear than specific query methods - they may be rmeoved
734
return self.has_id(file_id)
736
def id2path(self, file_id):
737
"""Return as a string the path to file_id.
740
>>> e = i.add(InventoryDirectory('src-id', 'src', ROOT_ID))
741
>>> e = i.add(InventoryFile('foo-id', 'foo.c', parent_id='src-id'))
742
>>> print i.id2path('foo-id')
745
# get all names, skipping root
746
return '/'.join(reversed(
747
[parent.name for parent in
748
self._iter_file_id_parents(file_id)][:-1]))
750
def iter_entries(self, from_dir=None, recursive=True):
751
"""Return (path, entry) pairs, in order by name.
753
:param from_dir: if None, start from the root,
754
otherwise start from this directory (either file-id or entry)
755
:param recursive: recurse into directories or not
758
if self.root is None:
762
elif isinstance(from_dir, basestring):
763
from_dir = self[from_dir]
765
# unrolling the recursive called changed the time from
766
# 440ms/663ms (inline/total) to 116ms/116ms
767
children = from_dir.children.items()
770
for name, ie in children:
773
children = collections.deque(children)
774
stack = [(u'', children)]
776
from_dir_relpath, children = stack[-1]
779
name, ie = children.popleft()
781
# we know that from_dir_relpath never ends in a slash
782
# and 'f' doesn't begin with one, we can do a string op, rather
783
# than the checks of pathjoin(), though this means that all paths
785
path = from_dir_relpath + '/' + name
789
if ie.kind != 'directory':
792
# But do this child first
793
new_children = ie.children.items()
795
new_children = collections.deque(new_children)
796
stack.append((path, new_children))
797
# Break out of inner loop, so that we start outer loop with child
800
# if we finished all children, pop it off the stack
803
def iter_entries_by_dir(self, from_dir=None, specific_file_ids=None,
804
yield_parents=False):
805
"""Iterate over the entries in a directory first order.
807
This returns all entries for a directory before returning
808
the entries for children of a directory. This is not
809
lexicographically sorted order, and is a hybrid between
810
depth-first and breadth-first.
812
:param yield_parents: If True, yield the parents from the root leading
813
down to specific_file_ids that have been requested. This has no
814
impact if specific_file_ids is None.
815
:return: This yields (path, entry) pairs
817
if specific_file_ids and not isinstance(specific_file_ids, set):
818
specific_file_ids = set(specific_file_ids)
819
# TODO? Perhaps this should return the from_dir so that the root is
820
# yielded? or maybe an option?
822
if self.root is None:
824
# Optimize a common case
825
if (not yield_parents and specific_file_ids is not None and
826
len(specific_file_ids) == 1):
827
file_id = list(specific_file_ids)[0]
829
yield self.id2path(file_id), self[file_id]
832
if (specific_file_ids is None or yield_parents or
833
self.root.file_id in specific_file_ids):
835
elif isinstance(from_dir, basestring):
836
from_dir = self[from_dir]
838
if specific_file_ids is not None:
839
# TODO: jam 20070302 This could really be done as a loop rather
840
# than a bunch of recursive calls.
843
def add_ancestors(file_id):
844
if file_id not in byid:
846
parent_id = byid[file_id].parent_id
847
if parent_id is None:
849
if parent_id not in parents:
850
parents.add(parent_id)
851
add_ancestors(parent_id)
852
for file_id in specific_file_ids:
853
add_ancestors(file_id)
857
stack = [(u'', from_dir)]
859
cur_relpath, cur_dir = stack.pop()
862
for child_name, child_ie in sorted(cur_dir.children.iteritems()):
864
child_relpath = cur_relpath + child_name
866
if (specific_file_ids is None or
867
child_ie.file_id in specific_file_ids or
868
(yield_parents and child_ie.file_id in parents)):
869
yield child_relpath, child_ie
871
if child_ie.kind == 'directory':
872
if parents is None or child_ie.file_id in parents:
873
child_dirs.append((child_relpath+'/', child_ie))
874
stack.extend(reversed(child_dirs))
876
def _make_delta(self, old):
877
"""Make an inventory delta from two inventories."""
880
adds = new_ids - old_ids
881
deletes = old_ids - new_ids
882
common = old_ids.intersection(new_ids)
884
for file_id in deletes:
885
delta.append((old.id2path(file_id), None, file_id, None))
887
delta.append((None, self.id2path(file_id), file_id, self[file_id]))
888
for file_id in common:
889
if old[file_id] != self[file_id]:
890
delta.append((old.id2path(file_id), self.id2path(file_id),
891
file_id, self[file_id]))
894
def _get_mutable_inventory(self):
895
"""Returns a mutable copy of the object.
897
Some inventories are immutable, yet working trees, for example, needs
898
to mutate exisiting inventories instead of creating a new one.
900
raise NotImplementedError(self._get_mutable_inventory)
902
def make_entry(self, kind, name, parent_id, file_id=None):
903
"""Simple thunk to bzrlib.inventory.make_entry."""
904
return make_entry(kind, name, parent_id, file_id)
907
"""Return list of (path, ie) for all entries except the root.
909
This may be faster than iter_entries.
912
def descend(dir_ie, dir_path):
913
kids = dir_ie.children.items()
915
for name, ie in kids:
916
child_path = osutils.pathjoin(dir_path, name)
917
accum.append((child_path, ie))
918
if ie.kind == 'directory':
919
descend(ie, child_path)
921
descend(self.root, u'')
924
def directories(self):
925
"""Return (path, entry) pairs for all directories, including the root.
928
def descend(parent_ie, parent_path):
929
accum.append((parent_path, parent_ie))
931
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
934
for name, child_ie in kids:
935
child_path = osutils.pathjoin(parent_path, name)
936
descend(child_ie, child_path)
937
descend(self.root, u'')
940
def path2id(self, name):
941
"""Walk down through directories to return entry of last component.
943
names may be either a list of path components, or a single
944
string, in which case it is automatically split.
946
This returns the entry of the last component in the path,
947
which may be either a file or a directory.
949
Returns None IFF the path is not found.
951
if isinstance(name, basestring):
952
name = osutils.splitpath(name)
954
# mutter("lookup path %r" % name)
958
except errors.NoSuchId:
959
# root doesn't exist yet so nothing else can
965
children = getattr(parent, 'children', None)
974
return parent.file_id
976
def filter(self, specific_fileids):
977
"""Get an inventory view filtered against a set of file-ids.
979
Children of directories and parents are included.
981
The result may or may not reference the underlying inventory
982
so it should be treated as immutable.
984
interesting_parents = set()
985
for fileid in specific_fileids:
987
interesting_parents.update(self.get_idpath(fileid))
988
except errors.NoSuchId:
989
# This fileid is not in the inventory - that's ok
991
entries = self.iter_entries()
992
if self.root is None:
993
return Inventory(root_id=None)
994
other = Inventory(entries.next()[1].file_id)
995
other.root.revision = self.root.revision
996
other.revision_id = self.revision_id
997
directories_to_expand = set()
998
for path, entry in entries:
999
file_id = entry.file_id
1000
if (file_id in specific_fileids
1001
or entry.parent_id in directories_to_expand):
1002
if entry.kind == 'directory':
1003
directories_to_expand.add(file_id)
1004
elif file_id not in interesting_parents:
1006
other.add(entry.copy())
1009
def get_idpath(self, file_id):
1010
"""Return a list of file_ids for the path to an entry.
1012
The list contains one element for each directory followed by
1013
the id of the file itself. So the length of the returned list
1014
is equal to the depth of the file in the tree, counting the
1015
root directory as depth 1.
1018
for parent in self._iter_file_id_parents(file_id):
1019
p.insert(0, parent.file_id)
1023
class Inventory(CommonInventory):
1024
"""Inventory of versioned files in a tree.
1026
This describes which file_id is present at each point in the tree,
1027
and possibly the SHA-1 or other information about the file.
1028
Entries can be looked up either by path or by file_id.
1030
The inventory represents a typical unix file tree, with
1031
directories containing files and subdirectories. We never store
1032
the full path to a file, because renaming a directory implicitly
1033
moves all of its contents. This class internally maintains a
1034
lookup tree that allows the children under a directory to be
1037
InventoryEntry objects must not be modified after they are
1038
inserted, other than through the Inventory API.
1040
>>> inv = Inventory()
1041
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
1042
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
1043
>>> inv['123-123'].name
1046
May be treated as an iterator or set to look up file ids:
1048
>>> bool(inv.path2id('hello.c'))
1050
>>> '123-123' in inv
1053
May also look up by name:
1055
>>> [x[0] for x in inv.iter_entries()]
1057
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
1058
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
1059
Traceback (most recent call last):
1060
BzrError: parent_id {TREE_ROOT} not in inventory
1061
>>> inv.add(InventoryFile('123-123', 'hello.c', 'TREE_ROOT-12345678-12345678'))
1062
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT-12345678-12345678', sha1=None, len=None, revision=None)
1064
def __init__(self, root_id=ROOT_ID, revision_id=None):
1065
"""Create or read an inventory.
1067
If a working directory is specified, the inventory is read
1068
from there. If the file is specified, read from that. If not,
1069
the inventory is created empty.
1071
The inventory is created with a default root directory, with
1074
if root_id is not None:
1075
self._set_root(InventoryDirectory(root_id, u'', None))
1079
self.revision_id = revision_id
1082
# More than one page of ouput is not useful anymore to debug
1085
contents = repr(self._byid)
1086
if len(contents) > max_len:
1087
contents = contents[:(max_len-len(closing))] + closing
1088
return "<Inventory object at %x, contents=%r>" % (id(self), contents)
1090
def apply_delta(self, delta):
1091
"""Apply a delta to this inventory.
1093
:param delta: A list of changes to apply. After all the changes are
1094
applied the final inventory must be internally consistent, but it
1095
is ok to supply changes which, if only half-applied would have an
1096
invalid result - such as supplying two changes which rename two
1097
files, 'A' and 'B' with each other : [('A', 'B', 'A-id', a_entry),
1098
('B', 'A', 'B-id', b_entry)].
1100
Each change is a tuple, of the form (old_path, new_path, file_id,
1103
When new_path is None, the change indicates the removal of an entry
1104
from the inventory and new_entry will be ignored (using None is
1105
appropriate). If new_path is not None, then new_entry must be an
1106
InventoryEntry instance, which will be incorporated into the
1107
inventory (and replace any existing entry with the same file id).
1109
When old_path is None, the change indicates the addition of
1110
a new entry to the inventory.
1112
When neither new_path nor old_path are None, the change is a
1113
modification to an entry, such as a rename, reparent, kind change
1116
The children attribute of new_entry is ignored. This is because
1117
this method preserves children automatically across alterations to
1118
the parent of the children, and cases where the parent id of a
1119
child is changing require the child to be passed in as a separate
1120
change regardless. E.g. in the recursive deletion of a directory -
1121
the directory's children must be included in the delta, or the
1122
final inventory will be invalid.
1124
Note that a file_id must only appear once within a given delta.
1125
An AssertionError is raised otherwise.
1127
# Check that the delta is legal. It would be nice if this could be
1128
# done within the loops below but it's safer to validate the delta
1129
# before starting to mutate the inventory.
1130
unique_file_ids = set([f for _, _, f, _ in delta])
1131
if len(unique_file_ids) != len(delta):
1132
raise AssertionError("a file-id appears multiple times in %r"
1137
# Remove all affected items which were in the original inventory,
1138
# starting with the longest paths, thus ensuring parents are examined
1139
# after their children, which means that everything we examine has no
1140
# modified children remaining by the time we examine it.
1141
for old_path, file_id in sorted(((op, f) for op, np, f, e in delta
1142
if op is not None), reverse=True):
1143
if file_id not in self:
1146
# Preserve unaltered children of file_id for later reinsertion.
1147
file_id_children = getattr(self[file_id], 'children', {})
1148
if len(file_id_children):
1149
children[file_id] = file_id_children
1150
# Remove file_id and the unaltered children. If file_id is not
1151
# being deleted it will be reinserted back later.
1152
self.remove_recursive_id(file_id)
1153
# Insert all affected which should be in the new inventory, reattaching
1154
# their children if they had any. This is done from shortest path to
1155
# longest, ensuring that items which were modified and whose parents in
1156
# the resulting inventory were also modified, are inserted after their
1158
for new_path, new_entry in sorted((np, e) for op, np, f, e in
1159
delta if np is not None):
1160
if new_entry.kind == 'directory':
1161
# Pop the child which to allow detection of children whose
1162
# parents were deleted and which were not reattached to a new
1164
replacement = InventoryDirectory(new_entry.file_id,
1165
new_entry.name, new_entry.parent_id)
1166
replacement.revision = new_entry.revision
1167
replacement.children = children.pop(replacement.file_id, {})
1168
new_entry = replacement
1171
# Get the parent id that was deleted
1172
parent_id, children = children.popitem()
1173
raise errors.InconsistentDelta("<deleted>", parent_id,
1174
"The file id was deleted but its children were not deleted.")
1176
def _set_root(self, ie):
1178
self._byid = {self.root.file_id: self.root}
1181
# TODO: jam 20051218 Should copy also copy the revision_id?
1182
entries = self.iter_entries()
1183
if self.root is None:
1184
return Inventory(root_id=None)
1185
other = Inventory(entries.next()[1].file_id)
1186
other.root.revision = self.root.revision
1187
# copy recursively so we know directories will be added before
1188
# their children. There are more efficient ways than this...
1189
for path, entry in entries:
1190
other.add(entry.copy())
1193
def _get_mutable_inventory(self):
1194
"""See CommonInventory._get_mutable_inventory."""
1195
return deepcopy(self)
1198
"""Iterate over all file-ids."""
1199
return iter(self._byid)
1201
def iter_just_entries(self):
1202
"""Iterate over all entries.
1204
Unlike iter_entries(), just the entries are returned (not (path, ie))
1205
and the order of entries is undefined.
1207
XXX: We may not want to merge this into bzr.dev.
1209
if self.root is None:
1211
for _, ie in self._byid.iteritems():
1215
"""Returns number of entries."""
1216
return len(self._byid)
1218
def __getitem__(self, file_id):
1219
"""Return the entry for given file_id.
1221
>>> inv = Inventory()
1222
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
1223
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
1224
>>> inv['123123'].name
1228
return self._byid[file_id]
1230
# really we're passing an inventory, not a tree...
1231
raise errors.NoSuchId(self, file_id)
1233
def get_file_kind(self, file_id):
1234
return self._byid[file_id].kind
1236
def get_child(self, parent_id, filename):
1237
return self[parent_id].children.get(filename)
1239
def _add_child(self, entry):
1240
"""Add an entry to the inventory, without adding it to its parent"""
1241
if entry.file_id in self._byid:
1242
raise BzrError("inventory already contains entry with id {%s}" %
1244
self._byid[entry.file_id] = entry
1245
for child in getattr(entry, 'children', {}).itervalues():
1246
self._add_child(child)
1249
def add(self, entry):
1250
"""Add entry to inventory.
1252
To add a file to a branch ready to be committed, use Branch.add,
1255
Returns the new entry object.
1257
if entry.file_id in self._byid:
1258
raise errors.DuplicateFileId(entry.file_id,
1259
self._byid[entry.file_id])
1261
if entry.parent_id is None:
1265
parent = self._byid[entry.parent_id]
1267
raise BzrError("parent_id {%s} not in inventory" %
1270
if entry.name in parent.children:
1271
raise BzrError("%s is already versioned" %
1272
osutils.pathjoin(self.id2path(parent.file_id),
1273
entry.name).encode('utf-8'))
1274
parent.children[entry.name] = entry
1275
return self._add_child(entry)
1277
def add_path(self, relpath, kind, file_id=None, parent_id=None):
1278
"""Add entry from a path.
1280
The immediate parent must already be versioned.
1282
Returns the new entry object."""
1284
parts = osutils.splitpath(relpath)
1288
file_id = generate_ids.gen_root_id()
1289
self.root = InventoryDirectory(file_id, '', None)
1290
self._byid = {self.root.file_id: self.root}
1293
parent_path = parts[:-1]
1294
parent_id = self.path2id(parent_path)
1295
if parent_id is None:
1296
raise errors.NotVersionedError(path=parent_path)
1297
ie = make_entry(kind, parts[-1], parent_id, file_id)
1300
def __delitem__(self, file_id):
1301
"""Remove entry by id.
1303
>>> inv = Inventory()
1304
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1305
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
1313
del self._byid[file_id]
1314
if ie.parent_id is not None:
1315
del self[ie.parent_id].children[ie.name]
1317
def __eq__(self, other):
1318
"""Compare two sets by comparing their contents.
1320
>>> i1 = Inventory()
1321
>>> i2 = Inventory()
1324
>>> i1.add(InventoryFile('123', 'foo', ROOT_ID))
1325
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
1328
>>> i2.add(InventoryFile('123', 'foo', ROOT_ID))
1329
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
1333
if not isinstance(other, Inventory):
1334
return NotImplemented
1336
return self._byid == other._byid
1338
def __ne__(self, other):
1339
return not self.__eq__(other)
1342
raise ValueError('not hashable')
1344
def _iter_file_id_parents(self, file_id):
1345
"""Yield the parents of file_id up to the root."""
1346
while file_id is not None:
1348
ie = self._byid[file_id]
1350
raise errors.NoSuchId(tree=None, file_id=file_id)
1352
file_id = ie.parent_id
1354
def has_filename(self, filename):
1355
return bool(self.path2id(filename))
1357
def has_id(self, file_id):
1358
return (file_id in self._byid)
1360
def _make_delta(self, old):
1361
"""Make an inventory delta from two inventories."""
1362
old_getter = getattr(old, '_byid', old)
1363
new_getter = self._byid
1364
old_ids = set(old_getter)
1365
new_ids = set(new_getter)
1366
adds = new_ids - old_ids
1367
deletes = old_ids - new_ids
1368
if not adds and not deletes:
1371
common = old_ids.intersection(new_ids)
1373
for file_id in deletes:
1374
delta.append((old.id2path(file_id), None, file_id, None))
1375
for file_id in adds:
1376
delta.append((None, self.id2path(file_id), file_id, self[file_id]))
1377
for file_id in common:
1378
new_ie = new_getter[file_id]
1379
old_ie = old_getter[file_id]
1380
# If xml_serializer returns the cached InventoryEntries (rather
1381
# than always doing .copy()), inlining the 'is' check saves 2.7M
1382
# calls to __eq__. Under lsprof this saves 20s => 6s.
1383
# It is a minor improvement without lsprof.
1384
if old_ie is new_ie or old_ie == new_ie:
1387
delta.append((old.id2path(file_id), self.id2path(file_id),
1391
def remove_recursive_id(self, file_id):
1392
"""Remove file_id, and children, from the inventory.
1394
:param file_id: A file_id to remove.
1396
to_find_delete = [self._byid[file_id]]
1398
while to_find_delete:
1399
ie = to_find_delete.pop()
1400
to_delete.append(ie.file_id)
1401
if ie.kind == 'directory':
1402
to_find_delete.extend(ie.children.values())
1403
for file_id in reversed(to_delete):
1405
del self._byid[file_id]
1406
if ie.parent_id is not None:
1407
del self[ie.parent_id].children[ie.name]
1411
def rename(self, file_id, new_parent_id, new_name):
1412
"""Move a file within the inventory.
1414
This can change either the name, or the parent, or both.
1416
This does not move the working file.
1418
new_name = ensure_normalized_name(new_name)
1419
if not is_valid_name(new_name):
1420
raise BzrError("not an acceptable filename: %r" % new_name)
1422
new_parent = self._byid[new_parent_id]
1423
if new_name in new_parent.children:
1424
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
1426
new_parent_idpath = self.get_idpath(new_parent_id)
1427
if file_id in new_parent_idpath:
1428
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
1429
% (self.id2path(file_id), self.id2path(new_parent_id)))
1431
file_ie = self._byid[file_id]
1432
old_parent = self._byid[file_ie.parent_id]
1434
# TODO: Don't leave things messed up if this fails
1436
del old_parent.children[file_ie.name]
1437
new_parent.children[new_name] = file_ie
1439
file_ie.name = new_name
1440
file_ie.parent_id = new_parent_id
1442
def is_root(self, file_id):
1443
return self.root is not None and file_id == self.root.file_id
1446
class CHKInventory(CommonInventory):
1447
"""An inventory persisted in a CHK store.
1449
By design, a CHKInventory is immutable so many of the methods
1450
supported by Inventory - add, rename, apply_delta, etc - are *not*
1451
supported. To create a new CHKInventory, use create_by_apply_delta()
1452
or from_inventory(), say.
1454
Internally, a CHKInventory has one or two CHKMaps:
1456
* id_to_entry - a map from (file_id,) => InventoryEntry as bytes
1457
* parent_id_basename_to_file_id - a map from (parent_id, basename_utf8)
1460
The second map is optional and not present in early CHkRepository's.
1462
No caching is performed: every method call or item access will perform
1463
requests to the storage layer. As such, keep references to objects you
1467
def __init__(self, search_key_name):
1468
CommonInventory.__init__(self)
1469
self._fileid_to_entry_cache = {}
1470
self._path_to_fileid_cache = {}
1471
self._search_key_name = search_key_name
1473
def __eq__(self, other):
1474
"""Compare two sets by comparing their contents."""
1475
if not isinstance(other, CHKInventory):
1476
return NotImplemented
1478
this_key = self.id_to_entry.key()
1479
other_key = other.id_to_entry.key()
1480
this_pid_key = self.parent_id_basename_to_file_id.key()
1481
other_pid_key = other.parent_id_basename_to_file_id.key()
1482
if None in (this_key, this_pid_key, other_key, other_pid_key):
1484
return this_key == other_key and this_pid_key == other_pid_key
1486
def _entry_to_bytes(self, entry):
1487
"""Serialise entry as a single bytestring.
1489
:param Entry: An inventory entry.
1490
:return: A bytestring for the entry.
1493
ENTRY ::= FILE | DIR | SYMLINK | TREE
1494
FILE ::= "file: " COMMON SEP SHA SEP SIZE SEP EXECUTABLE
1495
DIR ::= "dir: " COMMON
1496
SYMLINK ::= "symlink: " COMMON SEP TARGET_UTF8
1497
TREE ::= "tree: " COMMON REFERENCE_REVISION
1498
COMMON ::= FILE_ID SEP PARENT_ID SEP NAME_UTF8 SEP REVISION
1501
if entry.parent_id is not None:
1502
parent_str = entry.parent_id
1505
name_str = entry.name.encode("utf8")
1506
if entry.kind == 'file':
1507
if entry.executable:
1511
return "file: %s\n%s\n%s\n%s\n%s\n%d\n%s" % (
1512
entry.file_id, parent_str, name_str, entry.revision,
1513
entry.text_sha1, entry.text_size, exec_str)
1514
elif entry.kind == 'directory':
1515
return "dir: %s\n%s\n%s\n%s" % (
1516
entry.file_id, parent_str, name_str, entry.revision)
1517
elif entry.kind == 'symlink':
1518
return "symlink: %s\n%s\n%s\n%s\n%s" % (
1519
entry.file_id, parent_str, name_str, entry.revision,
1520
entry.symlink_target.encode("utf8"))
1521
elif entry.kind == 'tree-reference':
1522
return "tree: %s\n%s\n%s\n%s\n%s" % (
1523
entry.file_id, parent_str, name_str, entry.revision,
1524
entry.reference_revision)
1526
raise ValueError("unknown kind %r" % entry.kind)
1529
def _bytes_to_utf8name_key(bytes):
1530
"""Get the file_id, revision_id key out of bytes."""
1531
# We don't normally care about name, except for times when we want
1532
# to filter out empty names because of non rich-root...
1533
sections = bytes.split('\n')
1534
kind, file_id = sections[0].split(': ')
1535
return (sections[2], file_id, sections[3])
1537
def _bytes_to_entry(self, bytes):
1538
"""Deserialise a serialised entry."""
1539
sections = bytes.split('\n')
1540
if sections[0].startswith("file: "):
1541
result = InventoryFile(sections[0][6:],
1542
sections[2].decode('utf8'),
1544
result.text_sha1 = sections[4]
1545
result.text_size = int(sections[5])
1546
result.executable = sections[6] == "Y"
1547
elif sections[0].startswith("dir: "):
1548
result = CHKInventoryDirectory(sections[0][5:],
1549
sections[2].decode('utf8'),
1551
elif sections[0].startswith("symlink: "):
1552
result = InventoryLink(sections[0][9:],
1553
sections[2].decode('utf8'),
1555
result.symlink_target = sections[4].decode('utf8')
1556
elif sections[0].startswith("tree: "):
1557
result = TreeReference(sections[0][6:],
1558
sections[2].decode('utf8'),
1560
result.reference_revision = sections[4]
1562
raise ValueError("Not a serialised entry %r" % bytes)
1563
result.revision = sections[3]
1564
if result.parent_id == '':
1565
result.parent_id = None
1566
self._fileid_to_entry_cache[result.file_id] = result
1569
def _get_mutable_inventory(self):
1570
"""See CommonInventory._get_mutable_inventory."""
1571
entries = self.iter_entries()
1572
inv = Inventory(None, self.revision_id)
1573
for path, inv_entry in entries:
1574
inv.add(inv_entry.copy())
1577
def create_by_apply_delta(self, inventory_delta, new_revision_id,
1578
propagate_caches=False):
1579
"""Create a new CHKInventory by applying inventory_delta to this one.
1581
:param inventory_delta: The inventory delta to apply. See
1582
Inventory.apply_delta for details.
1583
:param new_revision_id: The revision id of the resulting CHKInventory.
1584
:param propagate_caches: If True, the caches for this inventory are
1585
copied to and updated for the result.
1586
:return: The new CHKInventory.
1588
result = CHKInventory(self._search_key_name)
1589
if propagate_caches:
1590
# Just propagate the path-to-fileid cache for now
1591
result._path_to_fileid_cache = dict(self._path_to_fileid_cache.iteritems())
1592
search_key_func = chk_map.search_key_registry.get(self._search_key_name)
1593
self.id_to_entry._ensure_root()
1594
maximum_size = self.id_to_entry._root_node.maximum_size
1595
result.revision_id = new_revision_id
1596
result.id_to_entry = chk_map.CHKMap(
1597
self.id_to_entry._store,
1598
self.id_to_entry.key(),
1599
search_key_func=search_key_func)
1600
result.id_to_entry._ensure_root()
1601
result.id_to_entry._root_node.set_maximum_size(maximum_size)
1602
parent_id_basename_delta = []
1603
if self.parent_id_basename_to_file_id is not None:
1604
result.parent_id_basename_to_file_id = chk_map.CHKMap(
1605
self.parent_id_basename_to_file_id._store,
1606
self.parent_id_basename_to_file_id.key(),
1607
search_key_func=search_key_func)
1608
result.parent_id_basename_to_file_id._ensure_root()
1609
self.parent_id_basename_to_file_id._ensure_root()
1610
result_p_id_root = result.parent_id_basename_to_file_id._root_node
1611
p_id_root = self.parent_id_basename_to_file_id._root_node
1612
result_p_id_root.set_maximum_size(p_id_root.maximum_size)
1613
result_p_id_root._key_width = p_id_root._key_width
1615
result.parent_id_basename_to_file_id = None
1616
result.root_id = self.root_id
1617
id_to_entry_delta = []
1618
for old_path, new_path, file_id, entry in inventory_delta:
1621
result.root_id = file_id
1622
if new_path is None:
1627
if propagate_caches:
1629
del result._path_to_fileid_cache[old_path]
1633
new_key = (file_id,)
1634
new_value = result._entry_to_bytes(entry)
1635
# Update caches. It's worth doing this whether
1636
# we're propagating the old caches or not.
1637
result._path_to_fileid_cache[new_path] = file_id
1638
if old_path is None:
1641
old_key = (file_id,)
1642
id_to_entry_delta.append((old_key, new_key, new_value))
1643
if result.parent_id_basename_to_file_id is not None:
1644
# parent_id, basename changes
1645
if old_path is None:
1648
old_entry = self[file_id]
1649
old_key = self._parent_id_basename_key(old_entry)
1650
if new_path is None:
1654
new_key = self._parent_id_basename_key(entry)
1656
if old_key != new_key:
1657
# If the two keys are the same, the value will be unchanged
1658
# as its always the file id.
1659
parent_id_basename_delta.append((old_key, new_key, new_value))
1660
result.id_to_entry.apply_delta(id_to_entry_delta)
1661
if parent_id_basename_delta:
1662
result.parent_id_basename_to_file_id.apply_delta(parent_id_basename_delta)
1666
def deserialise(klass, chk_store, bytes, expected_revision_id):
1667
"""Deserialise a CHKInventory.
1669
:param chk_store: A CHK capable VersionedFiles instance.
1670
:param bytes: The serialised bytes.
1671
:param expected_revision_id: The revision ID we think this inventory is
1673
:return: A CHKInventory
1675
lines = bytes.split('\n')
1677
raise AssertionError('bytes to deserialize must end with an eol')
1679
if lines[0] != 'chkinventory:':
1680
raise ValueError("not a serialised CHKInventory: %r" % bytes)
1682
allowed_keys = frozenset(['root_id', 'revision_id', 'search_key_name',
1683
'parent_id_basename_to_file_id',
1685
for line in lines[1:]:
1686
key, value = line.split(': ', 1)
1687
if key not in allowed_keys:
1688
raise errors.BzrError('Unknown key in inventory: %r\n%r'
1691
raise errors.BzrError('Duplicate key in inventory: %r\n%r'
1694
revision_id = info['revision_id']
1695
root_id = info['root_id']
1696
search_key_name = info.get('search_key_name', 'plain')
1697
parent_id_basename_to_file_id = info.get(
1698
'parent_id_basename_to_file_id', None)
1699
id_to_entry = info['id_to_entry']
1701
result = CHKInventory(search_key_name)
1702
result.revision_id = revision_id
1703
result.root_id = root_id
1704
search_key_func = chk_map.search_key_registry.get(
1705
result._search_key_name)
1706
if parent_id_basename_to_file_id is not None:
1707
result.parent_id_basename_to_file_id = chk_map.CHKMap(
1708
chk_store, (parent_id_basename_to_file_id,),
1709
search_key_func=search_key_func)
1711
result.parent_id_basename_to_file_id = None
1713
result.id_to_entry = chk_map.CHKMap(chk_store, (id_to_entry,),
1714
search_key_func=search_key_func)
1715
if (result.revision_id,) != expected_revision_id:
1716
raise ValueError("Mismatched revision id and expected: %r, %r" %
1717
(result.revision_id, expected_revision_id))
1721
def from_inventory(klass, chk_store, inventory, maximum_size=0, search_key_name='plain'):
1722
"""Create a CHKInventory from an existing inventory.
1724
The content of inventory is copied into the chk_store, and a
1725
CHKInventory referencing that is returned.
1727
:param chk_store: A CHK capable VersionedFiles instance.
1728
:param inventory: The inventory to copy.
1729
:param maximum_size: The CHKMap node size limit.
1730
:param search_key_name: The identifier for the search key function
1732
result = klass(search_key_name)
1733
result.revision_id = inventory.revision_id
1734
result.root_id = inventory.root.file_id
1736
entry_to_bytes = result._entry_to_bytes
1737
parent_id_basename_key = result._parent_id_basename_key
1738
id_to_entry_dict = {}
1739
parent_id_basename_dict = {}
1740
for path, entry in inventory.iter_entries():
1741
id_to_entry_dict[(entry.file_id,)] = entry_to_bytes(entry)
1742
p_id_key = parent_id_basename_key(entry)
1743
parent_id_basename_dict[p_id_key] = entry.file_id
1745
result._populate_from_dicts(chk_store, id_to_entry_dict,
1746
parent_id_basename_dict, maximum_size=maximum_size)
1749
def _populate_from_dicts(self, chk_store, id_to_entry_dict,
1750
parent_id_basename_dict, maximum_size):
1751
search_key_func = chk_map.search_key_registry.get(self._search_key_name)
1752
root_key = chk_map.CHKMap.from_dict(chk_store, id_to_entry_dict,
1753
maximum_size=maximum_size, key_width=1,
1754
search_key_func=search_key_func)
1755
self.id_to_entry = chk_map.CHKMap(chk_store, root_key,
1757
root_key = chk_map.CHKMap.from_dict(chk_store,
1758
parent_id_basename_dict,
1759
maximum_size=maximum_size, key_width=2,
1760
search_key_func=search_key_func)
1761
self.parent_id_basename_to_file_id = chk_map.CHKMap(chk_store,
1762
root_key, search_key_func)
1764
def _parent_id_basename_key(self, entry):
1765
"""Create a key for a entry in a parent_id_basename_to_file_id index."""
1766
if entry.parent_id is not None:
1767
parent_id = entry.parent_id
1770
return parent_id, entry.name.encode('utf8')
1772
def __getitem__(self, file_id):
1773
"""map a single file_id -> InventoryEntry."""
1775
raise errors.NoSuchId(self, file_id)
1776
result = self._fileid_to_entry_cache.get(file_id, None)
1777
if result is not None:
1780
return self._bytes_to_entry(
1781
self.id_to_entry.iteritems([(file_id,)]).next()[1])
1782
except StopIteration:
1783
# really we're passing an inventory, not a tree...
1784
raise errors.NoSuchId(self, file_id)
1786
def has_id(self, file_id):
1787
# Perhaps have an explicit 'contains' method on CHKMap ?
1788
if self._fileid_to_entry_cache.get(file_id, None) is not None:
1790
return len(list(self.id_to_entry.iteritems([(file_id,)]))) == 1
1792
def is_root(self, file_id):
1793
return file_id == self.root_id
1795
def _iter_file_id_parents(self, file_id):
1796
"""Yield the parents of file_id up to the root."""
1797
while file_id is not None:
1801
raise errors.NoSuchId(tree=self, file_id=file_id)
1803
file_id = ie.parent_id
1806
"""Iterate over all file-ids."""
1807
for key, _ in self.id_to_entry.iteritems():
1810
def iter_just_entries(self):
1811
"""Iterate over all entries.
1813
Unlike iter_entries(), just the entries are returned (not (path, ie))
1814
and the order of entries is undefined.
1816
XXX: We may not want to merge this into bzr.dev.
1818
for key, entry in self.id_to_entry.iteritems():
1820
ie = self._fileid_to_entry_cache.get(file_id, None)
1822
ie = self._bytes_to_entry(entry)
1823
self._fileid_to_entry_cache[file_id] = ie
1826
def iter_changes(self, basis):
1827
"""Generate a Tree.iter_changes change list between this and basis.
1829
:param basis: Another CHKInventory.
1830
:return: An iterator over the changes between self and basis, as per
1831
tree.iter_changes().
1833
# We want: (file_id, (path_in_source, path_in_target),
1834
# changed_content, versioned, parent, name, kind,
1836
for key, basis_value, self_value in \
1837
self.id_to_entry.iter_changes(basis.id_to_entry):
1839
if basis_value is not None:
1840
basis_entry = basis._bytes_to_entry(basis_value)
1841
path_in_source = basis.id2path(file_id)
1842
basis_parent = basis_entry.parent_id
1843
basis_name = basis_entry.name
1844
basis_executable = basis_entry.executable
1846
path_in_source = None
1849
basis_executable = None
1850
if self_value is not None:
1851
self_entry = self._bytes_to_entry(self_value)
1852
path_in_target = self.id2path(file_id)
1853
self_parent = self_entry.parent_id
1854
self_name = self_entry.name
1855
self_executable = self_entry.executable
1857
path_in_target = None
1860
self_executable = None
1861
if basis_value is None:
1863
kind = (None, self_entry.kind)
1864
versioned = (False, True)
1865
elif self_value is None:
1867
kind = (basis_entry.kind, None)
1868
versioned = (True, False)
1870
kind = (basis_entry.kind, self_entry.kind)
1871
versioned = (True, True)
1872
changed_content = False
1873
if kind[0] != kind[1]:
1874
changed_content = True
1875
elif kind[0] == 'file':
1876
if (self_entry.text_size != basis_entry.text_size or
1877
self_entry.text_sha1 != basis_entry.text_sha1):
1878
changed_content = True
1879
elif kind[0] == 'symlink':
1880
if self_entry.symlink_target != basis_entry.symlink_target:
1881
changed_content = True
1882
elif kind[0] == 'tree-reference':
1883
if (self_entry.reference_revision !=
1884
basis_entry.reference_revision):
1885
changed_content = True
1886
parent = (basis_parent, self_parent)
1887
name = (basis_name, self_name)
1888
executable = (basis_executable, self_executable)
1889
if (not changed_content
1890
and parent[0] == parent[1]
1891
and name[0] == name[1]
1892
and executable[0] == executable[1]):
1893
# Could happen when only the revision changed for a directory
1896
yield (file_id, (path_in_source, path_in_target), changed_content,
1897
versioned, parent, name, kind, executable)
1900
"""Return the number of entries in the inventory."""
1901
return len(self.id_to_entry)
1903
def _make_delta(self, old):
1904
"""Make an inventory delta from two inventories."""
1905
if type(old) != CHKInventory:
1906
return CommonInventory._make_delta(self, old)
1908
for key, old_value, self_value in \
1909
self.id_to_entry.iter_changes(old.id_to_entry):
1911
if old_value is not None:
1912
old_path = old.id2path(file_id)
1915
if self_value is not None:
1916
entry = self._bytes_to_entry(self_value)
1917
self._fileid_to_entry_cache[file_id] = entry
1918
new_path = self.id2path(file_id)
1922
delta.append((old_path, new_path, file_id, entry))
1925
def path2id(self, name):
1926
"""See CommonInventory.path2id()."""
1927
result = self._path_to_fileid_cache.get(name, None)
1929
result = CommonInventory.path2id(self, name)
1930
self._path_to_fileid_cache[name] = result
1934
"""Serialise the inventory to lines."""
1935
lines = ["chkinventory:\n"]
1936
if self._search_key_name != 'plain':
1937
# custom ordering grouping things that don't change together
1938
lines.append('search_key_name: %s\n' % (self._search_key_name,))
1939
lines.append("root_id: %s\n" % self.root_id)
1940
lines.append('parent_id_basename_to_file_id: %s\n' %
1941
self.parent_id_basename_to_file_id.key())
1942
lines.append("revision_id: %s\n" % self.revision_id)
1943
lines.append("id_to_entry: %s\n" % self.id_to_entry.key())
1945
lines.append("revision_id: %s\n" % self.revision_id)
1946
lines.append("root_id: %s\n" % self.root_id)
1947
if self.parent_id_basename_to_file_id is not None:
1948
lines.append('parent_id_basename_to_file_id: %s\n' %
1949
self.parent_id_basename_to_file_id.key())
1950
lines.append("id_to_entry: %s\n" % self.id_to_entry.key())
1955
"""Get the root entry."""
1956
return self[self.root_id]
1959
class CHKInventoryDirectory(InventoryDirectory):
1960
"""A directory in an inventory."""
1962
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
1963
'text_id', 'parent_id', '_children', 'executable',
1964
'revision', 'symlink_target', 'reference_revision',
1967
def __init__(self, file_id, name, parent_id, chk_inventory):
1968
# Don't call InventoryDirectory.__init__ - it isn't right for this
1970
InventoryEntry.__init__(self, file_id, name, parent_id)
1971
self._children = None
1972
self.kind = 'directory'
1973
self._chk_inventory = chk_inventory
1977
"""Access the list of children of this directory.
1979
With a parent_id_basename_to_file_id index, loads all the children,
1980
without loads the entire index. Without is bad. A more sophisticated
1981
proxy object might be nice, to allow partial loading of children as
1982
well when specific names are accessed. (So path traversal can be
1983
written in the obvious way but not examine siblings.).
1985
if self._children is not None:
1986
return self._children
1987
# No longer supported
1988
if self._chk_inventory.parent_id_basename_to_file_id is None:
1989
raise AssertionError("Inventories without"
1990
" parent_id_basename_to_file_id are no longer supported")
1992
# XXX: Todo - use proxy objects for the children rather than loading
1993
# all when the attribute is referenced.
1994
parent_id_index = self._chk_inventory.parent_id_basename_to_file_id
1996
for (parent_id, name_utf8), file_id in parent_id_index.iteritems(
1997
key_filter=[(self.file_id,)]):
1998
child_keys.add((file_id,))
2000
for file_id_key in child_keys:
2001
entry = self._chk_inventory._fileid_to_entry_cache.get(
2002
file_id_key[0], None)
2003
if entry is not None:
2004
result[entry.name] = entry
2005
cached.add(file_id_key)
2006
child_keys.difference_update(cached)
2007
# populate; todo: do by name
2008
id_to_entry = self._chk_inventory.id_to_entry
2009
for file_id_key, bytes in id_to_entry.iteritems(child_keys):
2010
entry = self._chk_inventory._bytes_to_entry(bytes)
2011
result[entry.name] = entry
2012
self._chk_inventory._fileid_to_entry_cache[file_id_key[0]] = entry
2013
self._children = result
2017
'directory': InventoryDirectory,
2018
'file': InventoryFile,
2019
'symlink': InventoryLink,
2020
'tree-reference': TreeReference
2023
def make_entry(kind, name, parent_id, file_id=None):
2024
"""Create an inventory entry.
2026
:param kind: the type of inventory entry to create.
2027
:param name: the basename of the entry.
2028
:param parent_id: the parent_id of the entry.
2029
:param file_id: the file_id to use. if None, one will be created.
2032
file_id = generate_ids.gen_file_id(name)
2033
name = ensure_normalized_name(name)
2035
factory = entry_factory[kind]
2037
raise BzrError("unknown kind %r" % kind)
2038
return factory(file_id, name, parent_id)
2041
def ensure_normalized_name(name):
2044
:raises InvalidNormalization: When name is not normalized, and cannot be
2045
accessed on this platform by the normalized path.
2046
:return: The NFC normalised version of name.
2048
#------- This has been copied to bzrlib.dirstate.DirState.add, please
2049
# keep them synchronised.
2050
# we dont import normalized_filename directly because we want to be
2051
# able to change the implementation at runtime for tests.
2052
norm_name, can_access = osutils.normalized_filename(name)
2053
if norm_name != name:
2057
# TODO: jam 20060701 This would probably be more useful
2058
# if the error was raised with the full path
2059
raise errors.InvalidNormalization(name)
2065
def is_valid_name(name):
2067
if _NAME_RE is None:
2068
_NAME_RE = re.compile(r'^[^/\\]+$')
2070
return bool(_NAME_RE.match(name))
added
removed
bzrlib/inventory.py
