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_in, 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(deprecated_in((1, 6, 0)))
180
def get_tar_item(self, root, dp, now, tree):
181
"""Get a tarfile item and a file stream for its content."""
182
item = tarfile.TarInfo(osutils.pathjoin(root, dp).encode('utf8'))
183
# TODO: would be cool to actually set it to the timestamp of the
184
# revision it was last changed
186
fileobj = self._put_in_tar(item, tree)
190
"""Return true if the object this entry represents has textual data.
192
Note that textual data includes binary content.
194
Also note that all entries get weave files created for them.
195
This attribute is primarily used when upgrading from old trees that
196
did not have the weave index for all inventory entries.
200
def __init__(self, file_id, name, parent_id, text_id=None):
201
"""Create an InventoryEntry
203
The filename must be a single component, relative to the
204
parent directory; it cannot be a whole path or relative name.
206
>>> e = InventoryFile('123', 'hello.c', ROOT_ID)
211
>>> e = InventoryFile('123', 'src/hello.c', ROOT_ID)
212
Traceback (most recent call last):
213
InvalidEntryName: Invalid entry name: src/hello.c
215
if '/' in name or '\\' in name:
216
raise errors.InvalidEntryName(name=name)
217
self.executable = False
219
self.text_sha1 = None
220
self.text_size = None
221
self.file_id = file_id
223
self.text_id = text_id
224
self.parent_id = parent_id
225
self.symlink_target = None
226
self.reference_revision = None
228
def kind_character(self):
229
"""Return a short kind indicator useful for appending to names."""
230
raise BzrError('unknown kind %r' % self.kind)
232
known_kinds = ('file', 'directory', 'symlink')
234
def _put_in_tar(self, item, tree):
235
"""populate item for stashing in a tar, and return the content stream.
237
If no content is available, return None.
239
raise BzrError("don't know how to export {%s} of kind %r" %
240
(self.file_id, self.kind))
242
@deprecated_method(deprecated_in((1, 6, 0)))
243
def put_on_disk(self, dest, dp, tree):
244
"""Create a representation of self on disk in the prefix dest.
246
This is a template method - implement _put_on_disk in subclasses.
248
fullpath = osutils.pathjoin(dest, dp)
249
self._put_on_disk(fullpath, tree)
250
# mutter(" export {%s} kind %s to %s", self.file_id,
251
# self.kind, fullpath)
253
def _put_on_disk(self, fullpath, tree):
254
"""Put this entry onto disk at fullpath, from tree tree."""
255
raise BzrError("don't know how to export {%s} of kind %r" % (self.file_id, self.kind))
257
def sorted_children(self):
258
return sorted(self.children.items())
261
def versionable_kind(kind):
262
return (kind in ('file', 'directory', 'symlink', 'tree-reference'))
264
def check(self, checker, rev_id, inv, tree):
265
"""Check this inventory entry is intact.
267
This is a template method, override _check for kind specific
270
:param checker: Check object providing context for the checks;
271
can be used to find out what parts of the repository have already
273
:param rev_id: Revision id from which this InventoryEntry was loaded.
274
Not necessarily the last-changed revision for this file.
275
:param inv: Inventory from which the entry was loaded.
276
:param tree: RevisionTree for this entry.
278
if self.parent_id is not None:
279
if not inv.has_id(self.parent_id):
280
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
281
% (self.parent_id, rev_id))
282
self._check(checker, rev_id, tree)
284
def _check(self, checker, rev_id, tree):
285
"""Check this inventory entry for kind specific errors."""
286
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
290
"""Clone this inventory entry."""
291
raise NotImplementedError
294
def describe_change(old_entry, new_entry):
295
"""Describe the change between old_entry and this.
297
This smells of being an InterInventoryEntry situation, but as its
298
the first one, we're making it a static method for now.
300
An entry with a different parent, or different name is considered
301
to be renamed. Reparenting is an internal detail.
302
Note that renaming the parent does not trigger a rename for the
305
# TODO: Perhaps return an object rather than just a string
306
if old_entry is new_entry:
307
# also the case of both being None
309
elif old_entry is None:
311
elif new_entry is None:
313
if old_entry.kind != new_entry.kind:
315
text_modified, meta_modified = new_entry.detect_changes(old_entry)
316
if text_modified or meta_modified:
320
# TODO 20060511 (mbp, rbc) factor out 'detect_rename' here.
321
if old_entry.parent_id != new_entry.parent_id:
323
elif old_entry.name != new_entry.name:
327
if renamed and not modified:
328
return InventoryEntry.RENAMED
329
if modified and not renamed:
331
if modified and renamed:
332
return InventoryEntry.MODIFIED_AND_RENAMED
336
return ("%s(%r, %r, parent_id=%r, revision=%r)"
337
% (self.__class__.__name__,
343
def __eq__(self, other):
345
# For the case when objects are cached
347
if not isinstance(other, InventoryEntry):
348
return NotImplemented
350
return ((self.file_id == other.file_id)
351
and (self.name == other.name)
352
and (other.symlink_target == self.symlink_target)
353
and (self.text_sha1 == other.text_sha1)
354
and (self.text_size == other.text_size)
355
and (self.text_id == other.text_id)
356
and (self.parent_id == other.parent_id)
357
and (self.kind == other.kind)
358
and (self.revision == other.revision)
359
and (self.executable == other.executable)
360
and (self.reference_revision == other.reference_revision)
363
def __ne__(self, other):
364
return not (self == other)
367
raise ValueError('not hashable')
369
def _unchanged(self, previous_ie):
370
"""Has this entry changed relative to previous_ie.
372
This method should be overridden in child classes.
375
# different inv parent
376
if previous_ie.parent_id != self.parent_id:
379
elif previous_ie.name != self.name:
381
elif previous_ie.kind != self.kind:
385
def _read_tree_state(self, path, work_tree):
386
"""Populate fields in the inventory entry from the given tree.
388
Note that this should be modified to be a noop on virtual trees
389
as all entries created there are prepopulated.
391
# TODO: Rather than running this manually, we should check the
392
# working sha1 and other expensive properties when they're
393
# first requested, or preload them if they're already known
394
pass # nothing to do by default
396
def _forget_tree_state(self):
400
class RootEntry(InventoryEntry):
402
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
403
'text_id', 'parent_id', 'children', 'executable',
404
'revision', 'symlink_target', 'reference_revision']
406
def _check(self, checker, rev_id, tree):
407
"""See InventoryEntry._check"""
409
def __init__(self, file_id):
410
self.file_id = file_id
412
self.kind = 'directory'
413
self.parent_id = None
416
symbol_versioning.warn('RootEntry is deprecated as of bzr 0.10.'
417
' Please use InventoryDirectory instead.',
418
DeprecationWarning, stacklevel=2)
420
def __eq__(self, other):
421
if not isinstance(other, RootEntry):
422
return NotImplemented
424
return (self.file_id == other.file_id) \
425
and (self.children == other.children)
428
class InventoryDirectory(InventoryEntry):
429
"""A directory in an inventory."""
431
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
432
'text_id', 'parent_id', 'children', 'executable',
433
'revision', 'symlink_target', 'reference_revision']
435
def _check(self, checker, rev_id, tree):
436
"""See InventoryEntry._check"""
437
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
438
raise BzrCheckError('directory {%s} has text in revision {%s}'
439
% (self.file_id, rev_id))
442
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
443
other.revision = self.revision
444
# note that children are *not* copied; they're pulled across when
448
def __init__(self, file_id, name, parent_id):
449
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
451
self.kind = 'directory'
453
def kind_character(self):
454
"""See InventoryEntry.kind_character."""
457
def _put_in_tar(self, item, tree):
458
"""See InventoryEntry._put_in_tar."""
459
item.type = tarfile.DIRTYPE
466
def _put_on_disk(self, fullpath, tree):
467
"""See InventoryEntry._put_on_disk."""
471
class InventoryFile(InventoryEntry):
472
"""A file in an inventory."""
474
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
475
'text_id', 'parent_id', 'children', 'executable',
476
'revision', 'symlink_target', 'reference_revision']
478
def _check(self, checker, tree_revision_id, tree):
479
"""See InventoryEntry._check"""
480
key = (self.file_id, self.revision)
481
if key in checker.checked_texts:
482
prev_sha = checker.checked_texts[key]
483
if prev_sha != self.text_sha1:
485
'mismatched sha1 on {%s} in {%s} (%s != %s) %r' %
486
(self.file_id, tree_revision_id, prev_sha, self.text_sha1,
489
checker.repeated_text_cnt += 1
492
mutter('check version {%s} of {%s}', tree_revision_id, self.file_id)
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)"
571
% (self.__class__.__name__,
578
def _forget_tree_state(self):
579
self.text_sha1 = None
581
def _unchanged(self, previous_ie):
582
"""See InventoryEntry._unchanged."""
583
compatible = super(InventoryFile, self)._unchanged(previous_ie)
584
if self.text_sha1 != previous_ie.text_sha1:
587
# FIXME: 20050930 probe for the text size when getting sha1
588
# in _read_tree_state
589
self.text_size = previous_ie.text_size
590
if self.executable != previous_ie.executable:
595
class InventoryLink(InventoryEntry):
596
"""A file in an inventory."""
598
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
599
'text_id', 'parent_id', 'children', 'executable',
600
'revision', 'symlink_target', 'reference_revision']
602
def _check(self, checker, rev_id, tree):
603
"""See InventoryEntry._check"""
604
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
605
raise BzrCheckError('symlink {%s} has text in revision {%s}'
606
% (self.file_id, rev_id))
607
if self.symlink_target is None:
608
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
609
% (self.file_id, rev_id))
612
other = InventoryLink(self.file_id, self.name, self.parent_id)
613
other.symlink_target = self.symlink_target
614
other.revision = self.revision
617
def detect_changes(self, old_entry):
618
"""See InventoryEntry.detect_changes."""
619
# FIXME: which _modified field should we use ? RBC 20051003
620
text_modified = (self.symlink_target != old_entry.symlink_target)
622
mutter(" symlink target changed")
623
meta_modified = False
624
return text_modified, meta_modified
626
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
627
output_to, reverse=False):
628
"""See InventoryEntry._diff."""
629
from bzrlib.diff import DiffSymlink
630
old_target = self.symlink_target
631
if to_entry is not None:
632
new_target = to_entry.symlink_target
641
new_target, old_target = old_target, new_target
642
differ = DiffSymlink(old_tree, new_tree, output_to)
643
return differ.diff_symlink(old_target, new_target)
645
def __init__(self, file_id, name, parent_id):
646
super(InventoryLink, self).__init__(file_id, name, parent_id)
647
self.kind = 'symlink'
649
def kind_character(self):
650
"""See InventoryEntry.kind_character."""
653
def _put_in_tar(self, item, tree):
654
"""See InventoryEntry._put_in_tar."""
655
item.type = tarfile.SYMTYPE
659
item.linkname = self.symlink_target
662
def _put_on_disk(self, fullpath, tree):
663
"""See InventoryEntry._put_on_disk."""
665
os.symlink(self.symlink_target, fullpath)
667
raise BzrError("Failed to create symlink %r -> %r, error: %s" % (fullpath, self.symlink_target, e))
669
def _read_tree_state(self, path, work_tree):
670
"""See InventoryEntry._read_tree_state."""
671
self.symlink_target = work_tree.get_symlink_target(self.file_id)
673
def _forget_tree_state(self):
674
self.symlink_target = None
676
def _unchanged(self, previous_ie):
677
"""See InventoryEntry._unchanged."""
678
compatible = super(InventoryLink, self)._unchanged(previous_ie)
679
if self.symlink_target != previous_ie.symlink_target:
684
class TreeReference(InventoryEntry):
686
kind = 'tree-reference'
688
def __init__(self, file_id, name, parent_id, revision=None,
689
reference_revision=None):
690
InventoryEntry.__init__(self, file_id, name, parent_id)
691
self.revision = revision
692
self.reference_revision = reference_revision
695
return TreeReference(self.file_id, self.name, self.parent_id,
696
self.revision, self.reference_revision)
698
def _read_tree_state(self, path, work_tree):
699
"""Populate fields in the inventory entry from the given tree.
701
self.reference_revision = work_tree.get_reference_revision(
704
def _forget_tree_state(self):
705
self.reference_revision = None
707
def _unchanged(self, previous_ie):
708
"""See InventoryEntry._unchanged."""
709
compatible = super(TreeReference, self)._unchanged(previous_ie)
710
if self.reference_revision != previous_ie.reference_revision:
715
class Inventory(object):
716
"""Inventory of versioned files in a tree.
718
This describes which file_id is present at each point in the tree,
719
and possibly the SHA-1 or other information about the file.
720
Entries can be looked up either by path or by file_id.
722
The inventory represents a typical unix file tree, with
723
directories containing files and subdirectories. We never store
724
the full path to a file, because renaming a directory implicitly
725
moves all of its contents. This class internally maintains a
726
lookup tree that allows the children under a directory to be
729
InventoryEntry objects must not be modified after they are
730
inserted, other than through the Inventory API.
732
>>> inv = Inventory()
733
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
734
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
735
>>> inv['123-123'].name
738
May be treated as an iterator or set to look up file ids:
740
>>> bool(inv.path2id('hello.c'))
745
May also look up by name:
747
>>> [x[0] for x in inv.iter_entries()]
749
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
750
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
751
Traceback (most recent call last):
752
BzrError: parent_id {TREE_ROOT} not in inventory
753
>>> inv.add(InventoryFile('123-123', 'hello.c', 'TREE_ROOT-12345678-12345678'))
754
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT-12345678-12345678', sha1=None, len=None)
756
def __init__(self, root_id=ROOT_ID, revision_id=None):
757
"""Create or read an inventory.
759
If a working directory is specified, the inventory is read
760
from there. If the file is specified, read from that. If not,
761
the inventory is created empty.
763
The inventory is created with a default root directory, with
766
if root_id is not None:
767
self._set_root(InventoryDirectory(root_id, u'', None))
771
self.revision_id = revision_id
774
return "<Inventory object at %x, contents=%r>" % (id(self), self._byid)
776
def apply_delta(self, delta):
777
"""Apply a delta to this inventory.
779
:param delta: A list of changes to apply. After all the changes are
780
applied the final inventory must be internally consistent, but it
781
is ok to supply changes which, if only half-applied would have an
782
invalid result - such as supplying two changes which rename two
783
files, 'A' and 'B' with each other : [('A', 'B', 'A-id', a_entry),
784
('B', 'A', 'B-id', b_entry)].
786
Each change is a tuple, of the form (old_path, new_path, file_id,
789
When new_path is None, the change indicates the removal of an entry
790
from the inventory and new_entry will be ignored (using None is
791
appropriate). If new_path is not None, then new_entry must be an
792
InventoryEntry instance, which will be incorporated into the
793
inventory (and replace any existing entry with the same file id).
795
When old_path is None, the change indicates the addition of
796
a new entry to the inventory.
798
When neither new_path nor old_path are None, the change is a
799
modification to an entry, such as a rename, reparent, kind change
802
The children attribute of new_entry is ignored. This is because
803
this method preserves children automatically across alterations to
804
the parent of the children, and cases where the parent id of a
805
child is changing require the child to be passed in as a separate
806
change regardless. E.g. in the recursive deletion of a directory -
807
the directory's children must be included in the delta, or the
808
final inventory will be invalid.
811
# Remove all affected items which were in the original inventory,
812
# starting with the longest paths, thus ensuring parents are examined
813
# after their children, which means that everything we examine has no
814
# modified children remaining by the time we examine it.
815
for old_path, file_id in sorted(((op, f) for op, np, f, e in delta
816
if op is not None), reverse=True):
817
if file_id not in self:
820
# Preserve unaltered children of file_id for later reinsertion.
821
file_id_children = getattr(self[file_id], 'children', {})
822
if len(file_id_children):
823
children[file_id] = file_id_children
824
# Remove file_id and the unaltered children. If file_id is not
825
# being deleted it will be reinserted back later.
826
self.remove_recursive_id(file_id)
827
# Insert all affected which should be in the new inventory, reattaching
828
# their children if they had any. This is done from shortest path to
829
# longest, ensuring that items which were modified and whose parents in
830
# the resulting inventory were also modified, are inserted after their
832
for new_path, new_entry in sorted((np, e) for op, np, f, e in
833
delta if np is not None):
834
if new_entry.kind == 'directory':
835
# Pop the child which to allow detection of children whose
836
# parents were deleted and which were not reattached to a new
838
new_entry.children = children.pop(new_entry.file_id, {})
841
# Get the parent id that was deleted
842
parent_id, children = children.popitem()
843
raise errors.InconsistentDelta("<deleted>", parent_id,
844
"The file id was deleted but its children were not deleted.")
846
def _set_root(self, ie):
848
self._byid = {self.root.file_id: self.root}
851
# TODO: jam 20051218 Should copy also copy the revision_id?
852
entries = self.iter_entries()
853
if self.root is None:
854
return Inventory(root_id=None)
855
other = Inventory(entries.next()[1].file_id)
856
other.root.revision = self.root.revision
857
# copy recursively so we know directories will be added before
858
# their children. There are more efficient ways than this...
859
for path, entry in entries:
860
other.add(entry.copy())
864
return iter(self._byid)
867
"""Returns number of entries."""
868
return len(self._byid)
870
def iter_entries(self, from_dir=None):
871
"""Return (path, entry) pairs, in order by name."""
873
if self.root is None:
877
elif isinstance(from_dir, basestring):
878
from_dir = self._byid[from_dir]
880
# unrolling the recursive called changed the time from
881
# 440ms/663ms (inline/total) to 116ms/116ms
882
children = from_dir.children.items()
884
children = collections.deque(children)
885
stack = [(u'', children)]
887
from_dir_relpath, children = stack[-1]
890
name, ie = children.popleft()
892
# we know that from_dir_relpath never ends in a slash
893
# and 'f' doesn't begin with one, we can do a string op, rather
894
# than the checks of pathjoin(), though this means that all paths
896
path = from_dir_relpath + '/' + name
900
if ie.kind != 'directory':
903
# But do this child first
904
new_children = ie.children.items()
906
new_children = collections.deque(new_children)
907
stack.append((path, new_children))
908
# Break out of inner loop, so that we start outer loop with child
911
# if we finished all children, pop it off the stack
914
def iter_entries_by_dir(self, from_dir=None, specific_file_ids=None,
915
yield_parents=False):
916
"""Iterate over the entries in a directory first order.
918
This returns all entries for a directory before returning
919
the entries for children of a directory. This is not
920
lexicographically sorted order, and is a hybrid between
921
depth-first and breadth-first.
923
:param yield_parents: If True, yield the parents from the root leading
924
down to specific_file_ids that have been requested. This has no
925
impact if specific_file_ids is None.
926
:return: This yields (path, entry) pairs
928
if specific_file_ids and not isinstance(specific_file_ids, set):
929
specific_file_ids = set(specific_file_ids)
930
# TODO? Perhaps this should return the from_dir so that the root is
931
# yielded? or maybe an option?
933
if self.root is None:
935
# Optimize a common case
936
if (not yield_parents and specific_file_ids is not None and
937
len(specific_file_ids) == 1):
938
file_id = list(specific_file_ids)[0]
940
yield self.id2path(file_id), self[file_id]
943
if (specific_file_ids is None or yield_parents or
944
self.root.file_id in specific_file_ids):
946
elif isinstance(from_dir, basestring):
947
from_dir = self._byid[from_dir]
949
if specific_file_ids is not None:
950
# TODO: jam 20070302 This could really be done as a loop rather
951
# than a bunch of recursive calls.
954
def add_ancestors(file_id):
955
if file_id not in byid:
957
parent_id = byid[file_id].parent_id
958
if parent_id is None:
960
if parent_id not in parents:
961
parents.add(parent_id)
962
add_ancestors(parent_id)
963
for file_id in specific_file_ids:
964
add_ancestors(file_id)
968
stack = [(u'', from_dir)]
970
cur_relpath, cur_dir = stack.pop()
973
for child_name, child_ie in sorted(cur_dir.children.iteritems()):
975
child_relpath = cur_relpath + child_name
977
if (specific_file_ids is None or
978
child_ie.file_id in specific_file_ids or
979
(yield_parents and child_ie.file_id in parents)):
980
yield child_relpath, child_ie
982
if child_ie.kind == 'directory':
983
if parents is None or child_ie.file_id in parents:
984
child_dirs.append((child_relpath+'/', child_ie))
985
stack.extend(reversed(child_dirs))
987
def make_entry(self, kind, name, parent_id, file_id=None):
988
"""Simple thunk to bzrlib.inventory.make_entry."""
989
return make_entry(kind, name, parent_id, file_id)
992
"""Return list of (path, ie) for all entries except the root.
994
This may be faster than iter_entries.
997
def descend(dir_ie, dir_path):
998
kids = dir_ie.children.items()
1000
for name, ie in kids:
1001
child_path = osutils.pathjoin(dir_path, name)
1002
accum.append((child_path, ie))
1003
if ie.kind == 'directory':
1004
descend(ie, child_path)
1006
descend(self.root, u'')
1009
def directories(self):
1010
"""Return (path, entry) pairs for all directories, including the root.
1013
def descend(parent_ie, parent_path):
1014
accum.append((parent_path, parent_ie))
1016
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
1019
for name, child_ie in kids:
1020
child_path = osutils.pathjoin(parent_path, name)
1021
descend(child_ie, child_path)
1022
descend(self.root, u'')
1025
def __contains__(self, file_id):
1026
"""True if this entry contains a file with given id.
1028
>>> inv = Inventory()
1029
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1030
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
1036
return (file_id in self._byid)
1038
def __getitem__(self, file_id):
1039
"""Return the entry for given file_id.
1041
>>> inv = Inventory()
1042
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
1043
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
1044
>>> inv['123123'].name
1048
return self._byid[file_id]
1050
# really we're passing an inventory, not a tree...
1051
raise errors.NoSuchId(self, file_id)
1053
def get_file_kind(self, file_id):
1054
return self._byid[file_id].kind
1056
def get_child(self, parent_id, filename):
1057
return self[parent_id].children.get(filename)
1059
def _add_child(self, entry):
1060
"""Add an entry to the inventory, without adding it to its parent"""
1061
if entry.file_id in self._byid:
1062
raise BzrError("inventory already contains entry with id {%s}" %
1064
self._byid[entry.file_id] = entry
1065
for child in getattr(entry, 'children', {}).itervalues():
1066
self._add_child(child)
1069
def add(self, entry):
1070
"""Add entry to inventory.
1072
To add a file to a branch ready to be committed, use Branch.add,
1075
Returns the new entry object.
1077
if entry.file_id in self._byid:
1078
raise errors.DuplicateFileId(entry.file_id,
1079
self._byid[entry.file_id])
1081
if entry.parent_id is None:
1085
parent = self._byid[entry.parent_id]
1087
raise BzrError("parent_id {%s} not in inventory" %
1090
if entry.name in parent.children:
1091
raise BzrError("%s is already versioned" %
1092
osutils.pathjoin(self.id2path(parent.file_id),
1093
entry.name).encode('utf-8'))
1094
parent.children[entry.name] = entry
1095
return self._add_child(entry)
1097
def add_path(self, relpath, kind, file_id=None, parent_id=None):
1098
"""Add entry from a path.
1100
The immediate parent must already be versioned.
1102
Returns the new entry object."""
1104
parts = osutils.splitpath(relpath)
1108
file_id = generate_ids.gen_root_id()
1109
self.root = InventoryDirectory(file_id, '', None)
1110
self._byid = {self.root.file_id: self.root}
1113
parent_path = parts[:-1]
1114
parent_id = self.path2id(parent_path)
1115
if parent_id is None:
1116
raise errors.NotVersionedError(path=parent_path)
1117
ie = make_entry(kind, parts[-1], parent_id, file_id)
1120
def __delitem__(self, file_id):
1121
"""Remove entry by id.
1123
>>> inv = Inventory()
1124
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1125
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
1133
del self._byid[file_id]
1134
if ie.parent_id is not None:
1135
del self[ie.parent_id].children[ie.name]
1137
def __eq__(self, other):
1138
"""Compare two sets by comparing their contents.
1140
>>> i1 = Inventory()
1141
>>> i2 = Inventory()
1144
>>> i1.add(InventoryFile('123', 'foo', ROOT_ID))
1145
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
1148
>>> i2.add(InventoryFile('123', 'foo', ROOT_ID))
1149
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
1153
if not isinstance(other, Inventory):
1154
return NotImplemented
1156
return self._byid == other._byid
1158
def __ne__(self, other):
1159
return not self.__eq__(other)
1162
raise ValueError('not hashable')
1164
def _iter_file_id_parents(self, file_id):
1165
"""Yield the parents of file_id up to the root."""
1166
while file_id is not None:
1168
ie = self._byid[file_id]
1170
raise errors.NoSuchId(tree=None, file_id=file_id)
1172
file_id = ie.parent_id
1174
def get_idpath(self, file_id):
1175
"""Return a list of file_ids for the path to an entry.
1177
The list contains one element for each directory followed by
1178
the id of the file itself. So the length of the returned list
1179
is equal to the depth of the file in the tree, counting the
1180
root directory as depth 1.
1183
for parent in self._iter_file_id_parents(file_id):
1184
p.insert(0, parent.file_id)
1187
def id2path(self, file_id):
1188
"""Return as a string the path to file_id.
1191
>>> e = i.add(InventoryDirectory('src-id', 'src', ROOT_ID))
1192
>>> e = i.add(InventoryFile('foo-id', 'foo.c', parent_id='src-id'))
1193
>>> print i.id2path('foo-id')
1196
# get all names, skipping root
1197
return '/'.join(reversed(
1198
[parent.name for parent in
1199
self._iter_file_id_parents(file_id)][:-1]))
1201
def path2id(self, name):
1202
"""Walk down through directories to return entry of last component.
1204
names may be either a list of path components, or a single
1205
string, in which case it is automatically split.
1207
This returns the entry of the last component in the path,
1208
which may be either a file or a directory.
1210
Returns None IFF the path is not found.
1212
if isinstance(name, basestring):
1213
name = osutils.splitpath(name)
1215
# mutter("lookup path %r" % name)
1222
children = getattr(parent, 'children', None)
1223
if children is None:
1228
# or raise an error?
1231
return parent.file_id
1233
def has_filename(self, names):
1234
return bool(self.path2id(names))
1236
def has_id(self, file_id):
1237
return (file_id in self._byid)
1239
def _make_delta(self, old):
1240
"""Make an inventory delta from two inventories."""
1241
old_getter = getattr(old, '_byid', old)
1242
new_getter = self._byid
1243
old_ids = set(old_getter)
1244
new_ids = set(new_getter)
1245
adds = new_ids - old_ids
1246
deletes = old_ids - new_ids
1247
if not adds and not deletes:
1250
common = old_ids.intersection(new_ids)
1252
for file_id in deletes:
1253
delta.append((old.id2path(file_id), None, file_id, None))
1254
for file_id in adds:
1255
delta.append((None, self.id2path(file_id), file_id, self[file_id]))
1256
for file_id in common:
1257
new_ie = new_getter[file_id]
1258
old_ie = old_getter[file_id]
1259
# If xml_serializer returns the cached InventoryEntries (rather
1260
# than always doing .copy()), inlining the 'is' check saves 2.7M
1261
# calls to __eq__. Under lsprof this saves 20s => 6s.
1262
# It is a minor improvement without lsprof.
1263
if old_ie is new_ie or old_ie == new_ie:
1266
delta.append((old.id2path(file_id), self.id2path(file_id),
1270
def remove_recursive_id(self, file_id):
1271
"""Remove file_id, and children, from the inventory.
1273
:param file_id: A file_id to remove.
1275
to_find_delete = [self._byid[file_id]]
1277
while to_find_delete:
1278
ie = to_find_delete.pop()
1279
to_delete.append(ie.file_id)
1280
if ie.kind == 'directory':
1281
to_find_delete.extend(ie.children.values())
1282
for file_id in reversed(to_delete):
1284
del self._byid[file_id]
1285
if ie.parent_id is not None:
1286
del self[ie.parent_id].children[ie.name]
1290
def rename(self, file_id, new_parent_id, new_name):
1291
"""Move a file within the inventory.
1293
This can change either the name, or the parent, or both.
1295
This does not move the working file.
1297
new_name = ensure_normalized_name(new_name)
1298
if not is_valid_name(new_name):
1299
raise BzrError("not an acceptable filename: %r" % new_name)
1301
new_parent = self._byid[new_parent_id]
1302
if new_name in new_parent.children:
1303
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
1305
new_parent_idpath = self.get_idpath(new_parent_id)
1306
if file_id in new_parent_idpath:
1307
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
1308
% (self.id2path(file_id), self.id2path(new_parent_id)))
1310
file_ie = self._byid[file_id]
1311
old_parent = self._byid[file_ie.parent_id]
1313
# TODO: Don't leave things messed up if this fails
1315
del old_parent.children[file_ie.name]
1316
new_parent.children[new_name] = file_ie
1318
file_ie.name = new_name
1319
file_ie.parent_id = new_parent_id
1321
def is_root(self, file_id):
1322
return self.root is not None and file_id == self.root.file_id
1326
'directory': InventoryDirectory,
1327
'file': InventoryFile,
1328
'symlink': InventoryLink,
1329
'tree-reference': TreeReference
1332
def make_entry(kind, name, parent_id, file_id=None):
1333
"""Create an inventory entry.
1335
:param kind: the type of inventory entry to create.
1336
:param name: the basename of the entry.
1337
:param parent_id: the parent_id of the entry.
1338
:param file_id: the file_id to use. if None, one will be created.
1341
file_id = generate_ids.gen_file_id(name)
1342
name = ensure_normalized_name(name)
1344
factory = entry_factory[kind]
1346
raise BzrError("unknown kind %r" % kind)
1347
return factory(file_id, name, parent_id)
1350
def ensure_normalized_name(name):
1353
:raises InvalidNormalization: When name is not normalized, and cannot be
1354
accessed on this platform by the normalized path.
1355
:return: The NFC normalised version of name.
1357
#------- This has been copied to bzrlib.dirstate.DirState.add, please
1358
# keep them synchronised.
1359
# we dont import normalized_filename directly because we want to be
1360
# able to change the implementation at runtime for tests.
1361
norm_name, can_access = osutils.normalized_filename(name)
1362
if norm_name != name:
1366
# TODO: jam 20060701 This would probably be more useful
1367
# if the error was raised with the full path
1368
raise errors.InvalidNormalization(name)
1374
def is_valid_name(name):
1376
if _NAME_RE is None:
1377
_NAME_RE = re.compile(r'^[^/\\]+$')
1379
return bool(_NAME_RE.match(name))
added
removed
bzrlib/inventory.py
