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., 59 Temple Place, Suite 330, Boston, MA 02111-1307 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.
34
from bzrlib.lazy_import import lazy_import
35
lazy_import(globals(), """
49
from bzrlib.errors import (
53
from bzrlib.symbol_versioning import deprecated_method
54
from bzrlib.trace import mutter
57
class InventoryEntry(object):
58
"""Description of a versioned file.
60
An InventoryEntry has the following fields, which are also
61
present in the XML inventory-entry element:
66
(within the parent directory)
69
file_id of the parent directory, or ROOT_ID
72
the revision_id in which this variation of this file was
76
Indicates that this file should be executable on systems
80
sha-1 of the text of the file
83
size in bytes of the text of the file
85
(reading a version 4 tree created a text_id field.)
90
>>> i.add(InventoryDirectory('123', 'src', ROOT_ID))
91
InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None)
92
>>> i.add(InventoryFile('2323', 'hello.c', parent_id='123'))
93
InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None)
94
>>> shouldbe = {0: '', 1: 'src', 2: 'src/hello.c'}
95
>>> for ix, j in enumerate(i.iter_entries()):
96
... print (j[0] == shouldbe[ix], j[1])
98
(True, InventoryDirectory('TREE_ROOT', u'', parent_id=None, revision=None))
99
(True, InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None))
100
(True, InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None))
101
>>> i.add(InventoryFile('2324', 'bye.c', '123'))
102
InventoryFile('2324', 'bye.c', parent_id='123', sha1=None, len=None)
103
>>> i.add(InventoryDirectory('2325', 'wibble', '123'))
104
InventoryDirectory('2325', 'wibble', parent_id='123', revision=None)
105
>>> i.path2id('src/wibble')
109
>>> i.add(InventoryFile('2326', 'wibble.c', '2325'))
110
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
112
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
113
>>> for path, entry in i.iter_entries():
122
>>> i.id2path('2326')
123
'src/wibble/wibble.c'
126
# Constants returned by describe_change()
128
# TODO: These should probably move to some kind of FileChangeDescription
129
# class; that's like what's inside a TreeDelta but we want to be able to
130
# generate them just for one file at a time.
132
MODIFIED_AND_RENAMED = 'modified and renamed'
136
def detect_changes(self, old_entry):
137
"""Return a (text_modified, meta_modified) from this to old_entry.
139
_read_tree_state must have been called on self and old_entry prior to
140
calling detect_changes.
144
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
145
output_to, reverse=False):
146
"""Perform a diff between two entries of the same kind."""
148
def parent_candidates(self, previous_inventories):
149
"""Find possible per-file graph parents.
151
This is currently defined by:
152
- Select the last changed revision in the parent inventory.
153
- Do deal with a short lived bug in bzr 0.8's development two entries
154
that have the same last changed but different 'x' bit settings are
157
# revision:ie mapping for each ie found in previous_inventories.
159
# identify candidate head revision ids.
160
for inv in previous_inventories:
161
if self.file_id in inv:
162
ie = inv[self.file_id]
163
if ie.revision in candidates:
164
# same revision value in two different inventories:
165
# correct possible inconsistencies:
166
# * there was a bug in revision updates with 'x' bit
169
if candidates[ie.revision].executable != ie.executable:
170
candidates[ie.revision].executable = False
171
ie.executable = False
172
except AttributeError:
175
# add this revision as a candidate.
176
candidates[ie.revision] = ie
179
@deprecated_method(symbol_versioning.zero_ninetyone)
180
def find_previous_heads(self, previous_inventories,
181
versioned_file_store,
184
"""Return the revisions and entries that directly precede this.
186
Returned as a map from revision to inventory entry.
188
This is a map containing the file revisions in all parents
189
for which the file exists, and its revision is not a parent of
190
any other. If the file is new, the set will be empty.
192
:param versioned_file_store: A store where ancestry data on this
193
file id can be queried.
194
:param transaction: The transaction that queries to the versioned
195
file store should be completed under.
196
:param entry_vf: The entry versioned file, if its already available.
198
candidates = self.parent_candidates(previous_inventories)
200
# revision:ie mapping with one revision for each head.
202
# common case optimisation
203
if len(candidates) == 1:
204
# if there is only one candidate revision found
205
# then we can avoid opening the versioned file to access ancestry:
206
# there cannot be any ancestors to eliminate when there is
207
# only one revision available.
210
# --- what follows is now encapsulated in repository.get_graph.heads(),
211
# but that is not accessible from here as we have no repository
212
# pointer. Note that the repository.get_graph.heads() call can return
213
# different results *at the moment* because of the kind-changing check
214
# we have in parent_candidates().
216
# eliminate ancestors amongst the available candidates:
217
# heads are those that are not an ancestor of any other candidate
218
# - this provides convergence at a per-file level.
219
def get_ancestors(weave, entry):
220
return set(weave.get_ancestry(entry.revision, topo_sorted=False))
221
# revision: ancestor list for each head
223
for ie in candidates.values():
224
# may be an ancestor of a known head:
225
already_present = 0 != len(
226
[head for head in heads
227
if ie.revision in head_ancestors[head]])
229
# an ancestor of an analyzed candidate.
231
# not an ancestor of a known head:
232
# load the versioned file for this file id if needed
234
entry_vf = versioned_file_store.get_weave_or_empty(
235
self.file_id, transaction)
236
ancestors = get_ancestors(entry_vf, ie)
237
# may knock something else out:
238
check_heads = list(heads.keys())
239
for head in check_heads:
240
if head in ancestors:
241
# this previously discovered 'head' is not
242
# really a head - its an ancestor of the newly
245
head_ancestors[ie.revision] = ancestors
246
heads[ie.revision] = ie
249
def get_tar_item(self, root, dp, now, tree):
250
"""Get a tarfile item and a file stream for its content."""
251
item = tarfile.TarInfo(osutils.pathjoin(root, dp).encode('utf8'))
252
# TODO: would be cool to actually set it to the timestamp of the
253
# revision it was last changed
255
fileobj = self._put_in_tar(item, tree)
259
"""Return true if the object this entry represents has textual data.
261
Note that textual data includes binary content.
263
Also note that all entries get weave files created for them.
264
This attribute is primarily used when upgrading from old trees that
265
did not have the weave index for all inventory entries.
269
def __init__(self, file_id, name, parent_id, text_id=None):
270
"""Create an InventoryEntry
272
The filename must be a single component, relative to the
273
parent directory; it cannot be a whole path or relative name.
275
>>> e = InventoryFile('123', 'hello.c', ROOT_ID)
280
>>> e = InventoryFile('123', 'src/hello.c', ROOT_ID)
281
Traceback (most recent call last):
282
InvalidEntryName: Invalid entry name: src/hello.c
284
if '/' in name or '\\' in name:
285
raise errors.InvalidEntryName(name=name)
286
self.executable = False
288
self.text_sha1 = None
289
self.text_size = None
290
self.file_id = file_id
292
self.text_id = text_id
293
self.parent_id = parent_id
294
self.symlink_target = None
295
self.reference_revision = None
297
def kind_character(self):
298
"""Return a short kind indicator useful for appending to names."""
299
raise BzrError('unknown kind %r' % self.kind)
301
known_kinds = ('file', 'directory', 'symlink')
303
def _put_in_tar(self, item, tree):
304
"""populate item for stashing in a tar, and return the content stream.
306
If no content is available, return None.
308
raise BzrError("don't know how to export {%s} of kind %r" %
309
(self.file_id, self.kind))
311
def put_on_disk(self, dest, dp, tree):
312
"""Create a representation of self on disk in the prefix dest.
314
This is a template method - implement _put_on_disk in subclasses.
316
fullpath = osutils.pathjoin(dest, dp)
317
self._put_on_disk(fullpath, tree)
318
# mutter(" export {%s} kind %s to %s", self.file_id,
319
# self.kind, fullpath)
321
def _put_on_disk(self, fullpath, tree):
322
"""Put this entry onto disk at fullpath, from tree tree."""
323
raise BzrError("don't know how to export {%s} of kind %r" % (self.file_id, self.kind))
325
def sorted_children(self):
326
return sorted(self.children.items())
329
def versionable_kind(kind):
330
return (kind in ('file', 'directory', 'symlink', 'tree-reference'))
332
def check(self, checker, rev_id, inv, tree):
333
"""Check this inventory entry is intact.
335
This is a template method, override _check for kind specific
338
:param checker: Check object providing context for the checks;
339
can be used to find out what parts of the repository have already
341
:param rev_id: Revision id from which this InventoryEntry was loaded.
342
Not necessarily the last-changed revision for this file.
343
:param inv: Inventory from which the entry was loaded.
344
:param tree: RevisionTree for this entry.
346
if self.parent_id is not None:
347
if not inv.has_id(self.parent_id):
348
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
349
% (self.parent_id, rev_id))
350
self._check(checker, rev_id, tree)
352
def _check(self, checker, rev_id, tree):
353
"""Check this inventory entry for kind specific errors."""
354
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
358
"""Clone this inventory entry."""
359
raise NotImplementedError
362
def describe_change(old_entry, new_entry):
363
"""Describe the change between old_entry and this.
365
This smells of being an InterInventoryEntry situation, but as its
366
the first one, we're making it a static method for now.
368
An entry with a different parent, or different name is considered
369
to be renamed. Reparenting is an internal detail.
370
Note that renaming the parent does not trigger a rename for the
373
# TODO: Perhaps return an object rather than just a string
374
if old_entry is new_entry:
375
# also the case of both being None
377
elif old_entry is None:
379
elif new_entry is None:
381
if old_entry.kind != new_entry.kind:
383
text_modified, meta_modified = new_entry.detect_changes(old_entry)
384
if text_modified or meta_modified:
388
# TODO 20060511 (mbp, rbc) factor out 'detect_rename' here.
389
if old_entry.parent_id != new_entry.parent_id:
391
elif old_entry.name != new_entry.name:
395
if renamed and not modified:
396
return InventoryEntry.RENAMED
397
if modified and not renamed:
399
if modified and renamed:
400
return InventoryEntry.MODIFIED_AND_RENAMED
404
return ("%s(%r, %r, parent_id=%r, revision=%r)"
405
% (self.__class__.__name__,
411
def __eq__(self, other):
412
if not isinstance(other, InventoryEntry):
413
return NotImplemented
415
return ((self.file_id == other.file_id)
416
and (self.name == other.name)
417
and (other.symlink_target == self.symlink_target)
418
and (self.text_sha1 == other.text_sha1)
419
and (self.text_size == other.text_size)
420
and (self.text_id == other.text_id)
421
and (self.parent_id == other.parent_id)
422
and (self.kind == other.kind)
423
and (self.revision == other.revision)
424
and (self.executable == other.executable)
425
and (self.reference_revision == other.reference_revision)
428
def __ne__(self, other):
429
return not (self == other)
432
raise ValueError('not hashable')
434
def _unchanged(self, previous_ie):
435
"""Has this entry changed relative to previous_ie.
437
This method should be overridden in child classes.
440
# different inv parent
441
if previous_ie.parent_id != self.parent_id:
444
elif previous_ie.name != self.name:
446
elif previous_ie.kind != self.kind:
450
def _read_tree_state(self, path, work_tree):
451
"""Populate fields in the inventory entry from the given tree.
453
Note that this should be modified to be a noop on virtual trees
454
as all entries created there are prepopulated.
456
# TODO: Rather than running this manually, we should check the
457
# working sha1 and other expensive properties when they're
458
# first requested, or preload them if they're already known
459
pass # nothing to do by default
461
def _forget_tree_state(self):
465
class RootEntry(InventoryEntry):
467
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
468
'text_id', 'parent_id', 'children', 'executable',
469
'revision', 'symlink_target', 'reference_revision']
471
def _check(self, checker, rev_id, tree):
472
"""See InventoryEntry._check"""
474
def __init__(self, file_id):
475
self.file_id = file_id
477
self.kind = 'directory'
478
self.parent_id = None
481
symbol_versioning.warn('RootEntry is deprecated as of bzr 0.10.'
482
' Please use InventoryDirectory instead.',
483
DeprecationWarning, stacklevel=2)
485
def __eq__(self, other):
486
if not isinstance(other, RootEntry):
487
return NotImplemented
489
return (self.file_id == other.file_id) \
490
and (self.children == other.children)
493
class InventoryDirectory(InventoryEntry):
494
"""A directory in an inventory."""
496
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
497
'text_id', 'parent_id', 'children', 'executable',
498
'revision', 'symlink_target', 'reference_revision']
500
def _check(self, checker, rev_id, tree):
501
"""See InventoryEntry._check"""
502
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
503
raise BzrCheckError('directory {%s} has text in revision {%s}'
504
% (self.file_id, rev_id))
507
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
508
other.revision = self.revision
509
# note that children are *not* copied; they're pulled across when
513
def __init__(self, file_id, name, parent_id):
514
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
516
self.kind = 'directory'
518
def kind_character(self):
519
"""See InventoryEntry.kind_character."""
522
def _put_in_tar(self, item, tree):
523
"""See InventoryEntry._put_in_tar."""
524
item.type = tarfile.DIRTYPE
531
def _put_on_disk(self, fullpath, tree):
532
"""See InventoryEntry._put_on_disk."""
536
class InventoryFile(InventoryEntry):
537
"""A file in an inventory."""
539
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
540
'text_id', 'parent_id', 'children', 'executable',
541
'revision', 'symlink_target', 'reference_revision']
543
def _check(self, checker, tree_revision_id, tree):
544
"""See InventoryEntry._check"""
545
t = (self.file_id, self.revision)
546
if t in checker.checked_texts:
547
prev_sha = checker.checked_texts[t]
548
if prev_sha != self.text_sha1:
550
'mismatched sha1 on {%s} in {%s} (%s != %s) %r' %
551
(self.file_id, tree_revision_id, prev_sha, self.text_sha1,
554
checker.repeated_text_cnt += 1
557
if self.file_id not in checker.checked_weaves:
558
mutter('check weave {%s}', self.file_id)
559
w = tree._get_weave(self.file_id)
560
# Not passing a progress bar, because it creates a new
561
# progress, which overwrites the current progress,
562
# and doesn't look nice
564
checker.checked_weaves[self.file_id] = True
566
w = tree._get_weave(self.file_id)
568
mutter('check version {%s} of {%s}', tree_revision_id, self.file_id)
569
checker.checked_text_cnt += 1
570
# We can't check the length, because Weave doesn't store that
571
# information, and the whole point of looking at the weave's
572
# sha1sum is that we don't have to extract the text.
573
if self.text_sha1 != w.get_sha1s([self.revision])[0]:
574
raise BzrCheckError('text {%s} version {%s} wrong sha1'
575
% (self.file_id, self.revision))
576
checker.checked_texts[t] = self.text_sha1
579
other = InventoryFile(self.file_id, self.name, self.parent_id)
580
other.executable = self.executable
581
other.text_id = self.text_id
582
other.text_sha1 = self.text_sha1
583
other.text_size = self.text_size
584
other.revision = self.revision
587
def detect_changes(self, old_entry):
588
"""See InventoryEntry.detect_changes."""
589
text_modified = (self.text_sha1 != old_entry.text_sha1)
590
meta_modified = (self.executable != old_entry.executable)
591
return text_modified, meta_modified
593
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
594
output_to, reverse=False):
595
"""See InventoryEntry._diff."""
596
from bzrlib.diff import DiffText
597
from_file_id = self.file_id
599
to_file_id = to_entry.file_id
603
to_file_id, from_file_id = from_file_id, to_file_id
604
tree, to_tree = to_tree, tree
605
from_label, to_label = to_label, from_label
606
differ = DiffText(tree, to_tree, output_to, 'utf-8', '', '',
608
return differ.diff_text(from_file_id, to_file_id, from_label, to_label)
611
"""See InventoryEntry.has_text."""
614
def __init__(self, file_id, name, parent_id):
615
super(InventoryFile, self).__init__(file_id, name, parent_id)
618
def kind_character(self):
619
"""See InventoryEntry.kind_character."""
622
def _put_in_tar(self, item, tree):
623
"""See InventoryEntry._put_in_tar."""
624
item.type = tarfile.REGTYPE
625
fileobj = tree.get_file(self.file_id)
626
item.size = self.text_size
627
if tree.is_executable(self.file_id):
633
def _put_on_disk(self, fullpath, tree):
634
"""See InventoryEntry._put_on_disk."""
635
osutils.pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
636
if tree.is_executable(self.file_id):
637
os.chmod(fullpath, 0755)
639
def _read_tree_state(self, path, work_tree):
640
"""See InventoryEntry._read_tree_state."""
641
self.text_sha1 = work_tree.get_file_sha1(self.file_id, path=path)
642
# FIXME: 20050930 probe for the text size when getting sha1
643
# in _read_tree_state
644
self.executable = work_tree.is_executable(self.file_id, path=path)
647
return ("%s(%r, %r, parent_id=%r, sha1=%r, len=%s)"
648
% (self.__class__.__name__,
655
def _forget_tree_state(self):
656
self.text_sha1 = None
658
def _unchanged(self, previous_ie):
659
"""See InventoryEntry._unchanged."""
660
compatible = super(InventoryFile, self)._unchanged(previous_ie)
661
if self.text_sha1 != previous_ie.text_sha1:
664
# FIXME: 20050930 probe for the text size when getting sha1
665
# in _read_tree_state
666
self.text_size = previous_ie.text_size
667
if self.executable != previous_ie.executable:
672
class InventoryLink(InventoryEntry):
673
"""A file in an inventory."""
675
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
676
'text_id', 'parent_id', 'children', 'executable',
677
'revision', 'symlink_target', 'reference_revision']
679
def _check(self, checker, rev_id, tree):
680
"""See InventoryEntry._check"""
681
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
682
raise BzrCheckError('symlink {%s} has text in revision {%s}'
683
% (self.file_id, rev_id))
684
if self.symlink_target is None:
685
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
686
% (self.file_id, rev_id))
689
other = InventoryLink(self.file_id, self.name, self.parent_id)
690
other.symlink_target = self.symlink_target
691
other.revision = self.revision
694
def detect_changes(self, old_entry):
695
"""See InventoryEntry.detect_changes."""
696
# FIXME: which _modified field should we use ? RBC 20051003
697
text_modified = (self.symlink_target != old_entry.symlink_target)
699
mutter(" symlink target changed")
700
meta_modified = False
701
return text_modified, meta_modified
703
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
704
output_to, reverse=False):
705
"""See InventoryEntry._diff."""
706
from bzrlib.diff import DiffSymlink
707
old_target = self.symlink_target
708
if to_entry is not None:
709
new_target = to_entry.symlink_target
718
new_target, old_target = old_target, new_target
719
differ = DiffSymlink(old_tree, new_tree, output_to)
720
return differ.diff_symlink(old_target, new_target)
722
def __init__(self, file_id, name, parent_id):
723
super(InventoryLink, self).__init__(file_id, name, parent_id)
724
self.kind = 'symlink'
726
def kind_character(self):
727
"""See InventoryEntry.kind_character."""
730
def _put_in_tar(self, item, tree):
731
"""See InventoryEntry._put_in_tar."""
732
item.type = tarfile.SYMTYPE
736
item.linkname = self.symlink_target
739
def _put_on_disk(self, fullpath, tree):
740
"""See InventoryEntry._put_on_disk."""
742
os.symlink(self.symlink_target, fullpath)
744
raise BzrError("Failed to create symlink %r -> %r, error: %s" % (fullpath, self.symlink_target, e))
746
def _read_tree_state(self, path, work_tree):
747
"""See InventoryEntry._read_tree_state."""
748
self.symlink_target = work_tree.get_symlink_target(self.file_id)
750
def _forget_tree_state(self):
751
self.symlink_target = None
753
def _unchanged(self, previous_ie):
754
"""See InventoryEntry._unchanged."""
755
compatible = super(InventoryLink, self)._unchanged(previous_ie)
756
if self.symlink_target != previous_ie.symlink_target:
761
class TreeReference(InventoryEntry):
763
kind = 'tree-reference'
765
def __init__(self, file_id, name, parent_id, revision=None,
766
reference_revision=None):
767
InventoryEntry.__init__(self, file_id, name, parent_id)
768
self.revision = revision
769
self.reference_revision = reference_revision
772
return TreeReference(self.file_id, self.name, self.parent_id,
773
self.revision, self.reference_revision)
775
def _read_tree_state(self, path, work_tree):
776
"""Populate fields in the inventory entry from the given tree.
778
self.reference_revision = work_tree.get_reference_revision(
781
def _forget_tree_state(self):
782
self.reference_revision = None
784
def _unchanged(self, previous_ie):
785
"""See InventoryEntry._unchanged."""
786
compatible = super(TreeReference, self)._unchanged(previous_ie)
787
if self.reference_revision != previous_ie.reference_revision:
792
class Inventory(object):
793
"""Inventory of versioned files in a tree.
795
This describes which file_id is present at each point in the tree,
796
and possibly the SHA-1 or other information about the file.
797
Entries can be looked up either by path or by file_id.
799
The inventory represents a typical unix file tree, with
800
directories containing files and subdirectories. We never store
801
the full path to a file, because renaming a directory implicitly
802
moves all of its contents. This class internally maintains a
803
lookup tree that allows the children under a directory to be
806
InventoryEntry objects must not be modified after they are
807
inserted, other than through the Inventory API.
809
>>> inv = Inventory()
810
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
811
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
812
>>> inv['123-123'].name
815
May be treated as an iterator or set to look up file ids:
817
>>> bool(inv.path2id('hello.c'))
822
May also look up by name:
824
>>> [x[0] for x in inv.iter_entries()]
826
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
827
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
828
Traceback (most recent call last):
829
BzrError: parent_id {TREE_ROOT} not in inventory
830
>>> inv.add(InventoryFile('123-123', 'hello.c', 'TREE_ROOT-12345678-12345678'))
831
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT-12345678-12345678', sha1=None, len=None)
833
def __init__(self, root_id=ROOT_ID, revision_id=None):
834
"""Create or read an inventory.
836
If a working directory is specified, the inventory is read
837
from there. If the file is specified, read from that. If not,
838
the inventory is created empty.
840
The inventory is created with a default root directory, with
843
if root_id is not None:
844
self._set_root(InventoryDirectory(root_id, u'', None))
848
self.revision_id = revision_id
851
return "<Inventory object at %x, contents=%r>" % (id(self), self._byid)
853
def apply_delta(self, delta):
854
"""Apply a delta to this inventory.
856
:param delta: A list of changes to apply. After all the changes are
857
applied the final inventory must be internally consistent, but it
858
is ok to supply changes which, if only half-applied would have an
859
invalid result - such as supplying two changes which rename two
860
files, 'A' and 'B' with each other : [('A', 'B', 'A-id', a_entry),
861
('B', 'A', 'B-id', b_entry)].
863
Each change is a tuple, of the form (old_path, new_path, file_id,
866
When new_path is None, the change indicates the removal of an entry
867
from the inventory and new_entry will be ignored (using None is
868
appropriate). If new_path is not None, then new_entry must be an
869
InventoryEntry instance, which will be incorporated into the
870
inventory (and replace any existing entry with the same file id).
872
When old_path is None, the change indicates the addition of
873
a new entry to the inventory.
875
When neither new_path nor old_path are None, the change is a
876
modification to an entry, such as a rename, reparent, kind change
879
The children attribute of new_entry is ignored. This is because
880
this method preserves children automatically across alterations to
881
the parent of the children, and cases where the parent id of a
882
child is changing require the child to be passed in as a separate
883
change regardless. E.g. in the recursive deletion of a directory -
884
the directory's children must be included in the delta, or the
885
final inventory will be invalid.
888
# Remove all affected items which were in the original inventory,
889
# starting with the longest paths, thus ensuring parents are examined
890
# after their children, which means that everything we examine has no
891
# modified children remaining by the time we examine it.
892
for old_path, file_id in sorted(((op, f) for op, np, f, e in delta
893
if op is not None), reverse=True):
894
if file_id not in self:
897
# Preserve unaltered children of file_id for later reinsertion.
898
children[file_id] = getattr(self[file_id], 'children', {})
899
# Remove file_id and the unaltered children. If file_id is not
900
# being deleted it will be reinserted back later.
901
self.remove_recursive_id(file_id)
902
# Insert all affected which should be in the new inventory, reattaching
903
# their children if they had any. This is done from shortest path to
904
# longest, ensuring that items which were modified and whose parents in
905
# the resulting inventory were also modified, are inserted after their
907
for new_path, new_entry in sorted((np, e) for op, np, f, e in
908
delta if np is not None):
909
if new_entry.kind == 'directory':
910
new_entry.children = children.get(new_entry.file_id, {})
913
def _set_root(self, ie):
915
self._byid = {self.root.file_id: self.root}
918
# TODO: jam 20051218 Should copy also copy the revision_id?
919
entries = self.iter_entries()
920
if self.root is None:
921
return Inventory(root_id=None)
922
other = Inventory(entries.next()[1].file_id)
923
# copy recursively so we know directories will be added before
924
# their children. There are more efficient ways than this...
925
for path, entry in entries:
926
other.add(entry.copy())
930
return iter(self._byid)
933
"""Returns number of entries."""
934
return len(self._byid)
936
def iter_entries(self, from_dir=None):
937
"""Return (path, entry) pairs, in order by name."""
939
if self.root is None:
943
elif isinstance(from_dir, basestring):
944
from_dir = self._byid[from_dir]
946
# unrolling the recursive called changed the time from
947
# 440ms/663ms (inline/total) to 116ms/116ms
948
children = from_dir.children.items()
950
children = collections.deque(children)
951
stack = [(u'', children)]
953
from_dir_relpath, children = stack[-1]
956
name, ie = children.popleft()
958
# we know that from_dir_relpath never ends in a slash
959
# and 'f' doesn't begin with one, we can do a string op, rather
960
# than the checks of pathjoin(), though this means that all paths
962
path = from_dir_relpath + '/' + name
966
if ie.kind != 'directory':
969
# But do this child first
970
new_children = ie.children.items()
972
new_children = collections.deque(new_children)
973
stack.append((path, new_children))
974
# Break out of inner loop, so that we start outer loop with child
977
# if we finished all children, pop it off the stack
980
def iter_entries_by_dir(self, from_dir=None, specific_file_ids=None,
981
yield_parents=False):
982
"""Iterate over the entries in a directory first order.
984
This returns all entries for a directory before returning
985
the entries for children of a directory. This is not
986
lexicographically sorted order, and is a hybrid between
987
depth-first and breadth-first.
989
:param yield_parents: If True, yield the parents from the root leading
990
down to specific_file_ids that have been requested. This has no
991
impact if specific_file_ids is None.
992
:return: This yields (path, entry) pairs
994
if specific_file_ids and not isinstance(specific_file_ids, set):
995
specific_file_ids = set(specific_file_ids)
996
# TODO? Perhaps this should return the from_dir so that the root is
997
# yielded? or maybe an option?
999
if self.root is None:
1001
# Optimize a common case
1002
if (not yield_parents and specific_file_ids is not None and
1003
len(specific_file_ids) == 1):
1004
file_id = list(specific_file_ids)[0]
1006
yield self.id2path(file_id), self[file_id]
1008
from_dir = self.root
1009
if (specific_file_ids is None or yield_parents or
1010
self.root.file_id in specific_file_ids):
1011
yield u'', self.root
1012
elif isinstance(from_dir, basestring):
1013
from_dir = self._byid[from_dir]
1015
if specific_file_ids is not None:
1016
# TODO: jam 20070302 This could really be done as a loop rather
1017
# than a bunch of recursive calls.
1020
def add_ancestors(file_id):
1021
if file_id not in byid:
1023
parent_id = byid[file_id].parent_id
1024
if parent_id is None:
1026
if parent_id not in parents:
1027
parents.add(parent_id)
1028
add_ancestors(parent_id)
1029
for file_id in specific_file_ids:
1030
add_ancestors(file_id)
1034
stack = [(u'', from_dir)]
1036
cur_relpath, cur_dir = stack.pop()
1039
for child_name, child_ie in sorted(cur_dir.children.iteritems()):
1041
child_relpath = cur_relpath + child_name
1043
if (specific_file_ids is None or
1044
child_ie.file_id in specific_file_ids or
1045
(yield_parents and child_ie.file_id in parents)):
1046
yield child_relpath, child_ie
1048
if child_ie.kind == 'directory':
1049
if parents is None or child_ie.file_id in parents:
1050
child_dirs.append((child_relpath+'/', child_ie))
1051
stack.extend(reversed(child_dirs))
1053
def make_entry(self, kind, name, parent_id, file_id=None):
1054
"""Simple thunk to bzrlib.inventory.make_entry."""
1055
return make_entry(kind, name, parent_id, file_id)
1058
"""Return list of (path, ie) for all entries except the root.
1060
This may be faster than iter_entries.
1063
def descend(dir_ie, dir_path):
1064
kids = dir_ie.children.items()
1066
for name, ie in kids:
1067
child_path = osutils.pathjoin(dir_path, name)
1068
accum.append((child_path, ie))
1069
if ie.kind == 'directory':
1070
descend(ie, child_path)
1072
descend(self.root, u'')
1075
def directories(self):
1076
"""Return (path, entry) pairs for all directories, including the root.
1079
def descend(parent_ie, parent_path):
1080
accum.append((parent_path, parent_ie))
1082
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
1085
for name, child_ie in kids:
1086
child_path = osutils.pathjoin(parent_path, name)
1087
descend(child_ie, child_path)
1088
descend(self.root, u'')
1091
def __contains__(self, file_id):
1092
"""True if this entry contains a file with given id.
1094
>>> inv = Inventory()
1095
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1096
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
1102
return (file_id in self._byid)
1104
def __getitem__(self, file_id):
1105
"""Return the entry for given file_id.
1107
>>> inv = Inventory()
1108
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
1109
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
1110
>>> inv['123123'].name
1114
return self._byid[file_id]
1116
# really we're passing an inventory, not a tree...
1117
raise errors.NoSuchId(self, file_id)
1119
def get_file_kind(self, file_id):
1120
return self._byid[file_id].kind
1122
def get_child(self, parent_id, filename):
1123
return self[parent_id].children.get(filename)
1125
def _add_child(self, entry):
1126
"""Add an entry to the inventory, without adding it to its parent"""
1127
if entry.file_id in self._byid:
1128
raise BzrError("inventory already contains entry with id {%s}" %
1130
self._byid[entry.file_id] = entry
1131
for child in getattr(entry, 'children', {}).itervalues():
1132
self._add_child(child)
1135
def add(self, entry):
1136
"""Add entry to inventory.
1138
To add a file to a branch ready to be committed, use Branch.add,
1141
Returns the new entry object.
1143
if entry.file_id in self._byid:
1144
raise errors.DuplicateFileId(entry.file_id,
1145
self._byid[entry.file_id])
1147
if entry.parent_id is None:
1151
parent = self._byid[entry.parent_id]
1153
raise BzrError("parent_id {%s} not in inventory" %
1156
if entry.name in parent.children:
1157
raise BzrError("%s is already versioned" %
1158
osutils.pathjoin(self.id2path(parent.file_id),
1159
entry.name).encode('utf-8'))
1160
parent.children[entry.name] = entry
1161
return self._add_child(entry)
1163
def add_path(self, relpath, kind, file_id=None, parent_id=None):
1164
"""Add entry from a path.
1166
The immediate parent must already be versioned.
1168
Returns the new entry object."""
1170
parts = osutils.splitpath(relpath)
1174
file_id = generate_ids.gen_root_id()
1175
self.root = InventoryDirectory(file_id, '', None)
1176
self._byid = {self.root.file_id: self.root}
1179
parent_path = parts[:-1]
1180
parent_id = self.path2id(parent_path)
1181
if parent_id is None:
1182
raise errors.NotVersionedError(path=parent_path)
1183
ie = make_entry(kind, parts[-1], parent_id, file_id)
1186
def __delitem__(self, file_id):
1187
"""Remove entry by id.
1189
>>> inv = Inventory()
1190
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1191
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
1199
del self._byid[file_id]
1200
if ie.parent_id is not None:
1201
del self[ie.parent_id].children[ie.name]
1203
def __eq__(self, other):
1204
"""Compare two sets by comparing their contents.
1206
>>> i1 = Inventory()
1207
>>> i2 = Inventory()
1210
>>> i1.add(InventoryFile('123', 'foo', ROOT_ID))
1211
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
1214
>>> i2.add(InventoryFile('123', 'foo', ROOT_ID))
1215
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
1219
if not isinstance(other, Inventory):
1220
return NotImplemented
1222
return self._byid == other._byid
1224
def __ne__(self, other):
1225
return not self.__eq__(other)
1228
raise ValueError('not hashable')
1230
def _iter_file_id_parents(self, file_id):
1231
"""Yield the parents of file_id up to the root."""
1232
while file_id is not None:
1234
ie = self._byid[file_id]
1236
raise errors.NoSuchId(tree=None, file_id=file_id)
1238
file_id = ie.parent_id
1240
def get_idpath(self, file_id):
1241
"""Return a list of file_ids for the path to an entry.
1243
The list contains one element for each directory followed by
1244
the id of the file itself. So the length of the returned list
1245
is equal to the depth of the file in the tree, counting the
1246
root directory as depth 1.
1249
for parent in self._iter_file_id_parents(file_id):
1250
p.insert(0, parent.file_id)
1253
def id2path(self, file_id):
1254
"""Return as a string the path to file_id.
1257
>>> e = i.add(InventoryDirectory('src-id', 'src', ROOT_ID))
1258
>>> e = i.add(InventoryFile('foo-id', 'foo.c', parent_id='src-id'))
1259
>>> print i.id2path('foo-id')
1262
# get all names, skipping root
1263
return '/'.join(reversed(
1264
[parent.name for parent in
1265
self._iter_file_id_parents(file_id)][:-1]))
1267
def path2id(self, name):
1268
"""Walk down through directories to return entry of last component.
1270
names may be either a list of path components, or a single
1271
string, in which case it is automatically split.
1273
This returns the entry of the last component in the path,
1274
which may be either a file or a directory.
1276
Returns None IFF the path is not found.
1278
if isinstance(name, basestring):
1279
name = osutils.splitpath(name)
1281
# mutter("lookup path %r" % name)
1288
children = getattr(parent, 'children', None)
1289
if children is None:
1294
# or raise an error?
1297
return parent.file_id
1299
def has_filename(self, names):
1300
return bool(self.path2id(names))
1302
def has_id(self, file_id):
1303
return (file_id in self._byid)
1305
def remove_recursive_id(self, file_id):
1306
"""Remove file_id, and children, from the inventory.
1308
:param file_id: A file_id to remove.
1310
to_find_delete = [self._byid[file_id]]
1312
while to_find_delete:
1313
ie = to_find_delete.pop()
1314
to_delete.append(ie.file_id)
1315
if ie.kind == 'directory':
1316
to_find_delete.extend(ie.children.values())
1317
for file_id in reversed(to_delete):
1319
del self._byid[file_id]
1320
if ie.parent_id is not None:
1321
del self[ie.parent_id].children[ie.name]
1325
def rename(self, file_id, new_parent_id, new_name):
1326
"""Move a file within the inventory.
1328
This can change either the name, or the parent, or both.
1330
This does not move the working file.
1332
new_name = ensure_normalized_name(new_name)
1333
if not is_valid_name(new_name):
1334
raise BzrError("not an acceptable filename: %r" % new_name)
1336
new_parent = self._byid[new_parent_id]
1337
if new_name in new_parent.children:
1338
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
1340
new_parent_idpath = self.get_idpath(new_parent_id)
1341
if file_id in new_parent_idpath:
1342
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
1343
% (self.id2path(file_id), self.id2path(new_parent_id)))
1345
file_ie = self._byid[file_id]
1346
old_parent = self._byid[file_ie.parent_id]
1348
# TODO: Don't leave things messed up if this fails
1350
del old_parent.children[file_ie.name]
1351
new_parent.children[new_name] = file_ie
1353
file_ie.name = new_name
1354
file_ie.parent_id = new_parent_id
1356
def is_root(self, file_id):
1357
return self.root is not None and file_id == self.root.file_id
1361
'directory': InventoryDirectory,
1362
'file': InventoryFile,
1363
'symlink': InventoryLink,
1364
'tree-reference': TreeReference
1367
def make_entry(kind, name, parent_id, file_id=None):
1368
"""Create an inventory entry.
1370
:param kind: the type of inventory entry to create.
1371
:param name: the basename of the entry.
1372
:param parent_id: the parent_id of the entry.
1373
:param file_id: the file_id to use. if None, one will be created.
1376
file_id = generate_ids.gen_file_id(name)
1377
name = ensure_normalized_name(name)
1379
factory = entry_factory[kind]
1381
raise BzrError("unknown kind %r" % kind)
1382
return factory(file_id, name, parent_id)
1385
def ensure_normalized_name(name):
1388
:raises InvalidNormalization: When name is not normalized, and cannot be
1389
accessed on this platform by the normalized path.
1390
:return: The NFC normalised version of name.
1392
#------- This has been copied to bzrlib.dirstate.DirState.add, please
1393
# keep them synchronised.
1394
# we dont import normalized_filename directly because we want to be
1395
# able to change the implementation at runtime for tests.
1396
norm_name, can_access = osutils.normalized_filename(name)
1397
if norm_name != name:
1401
# TODO: jam 20060701 This would probably be more useful
1402
# if the error was raised with the full path
1403
raise errors.InvalidNormalization(name)
1409
def is_valid_name(name):
1411
if _NAME_RE is None:
1412
_NAME_RE = re.compile(r'^[^/\\]+$')
1414
return bool(_NAME_RE.match(name))
added
removed
bzrlib/inventory.py
