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):
344
if not isinstance(other, InventoryEntry):
345
return NotImplemented
347
return ((self.file_id == other.file_id)
348
and (self.name == other.name)
349
and (other.symlink_target == self.symlink_target)
350
and (self.text_sha1 == other.text_sha1)
351
and (self.text_size == other.text_size)
352
and (self.text_id == other.text_id)
353
and (self.parent_id == other.parent_id)
354
and (self.kind == other.kind)
355
and (self.revision == other.revision)
356
and (self.executable == other.executable)
357
and (self.reference_revision == other.reference_revision)
360
def __ne__(self, other):
361
return not (self == other)
364
raise ValueError('not hashable')
366
def _unchanged(self, previous_ie):
367
"""Has this entry changed relative to previous_ie.
369
This method should be overridden in child classes.
372
# different inv parent
373
if previous_ie.parent_id != self.parent_id:
376
elif previous_ie.name != self.name:
378
elif previous_ie.kind != self.kind:
382
def _read_tree_state(self, path, work_tree):
383
"""Populate fields in the inventory entry from the given tree.
385
Note that this should be modified to be a noop on virtual trees
386
as all entries created there are prepopulated.
388
# TODO: Rather than running this manually, we should check the
389
# working sha1 and other expensive properties when they're
390
# first requested, or preload them if they're already known
391
pass # nothing to do by default
393
def _forget_tree_state(self):
397
class RootEntry(InventoryEntry):
399
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
400
'text_id', 'parent_id', 'children', 'executable',
401
'revision', 'symlink_target', 'reference_revision']
403
def _check(self, checker, rev_id, tree):
404
"""See InventoryEntry._check"""
406
def __init__(self, file_id):
407
self.file_id = file_id
409
self.kind = 'directory'
410
self.parent_id = None
413
symbol_versioning.warn('RootEntry is deprecated as of bzr 0.10.'
414
' Please use InventoryDirectory instead.',
415
DeprecationWarning, stacklevel=2)
417
def __eq__(self, other):
418
if not isinstance(other, RootEntry):
419
return NotImplemented
421
return (self.file_id == other.file_id) \
422
and (self.children == other.children)
425
class InventoryDirectory(InventoryEntry):
426
"""A directory in an inventory."""
428
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
429
'text_id', 'parent_id', 'children', 'executable',
430
'revision', 'symlink_target', 'reference_revision']
432
def _check(self, checker, rev_id, tree):
433
"""See InventoryEntry._check"""
434
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
435
raise BzrCheckError('directory {%s} has text in revision {%s}'
436
% (self.file_id, rev_id))
439
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
440
other.revision = self.revision
441
# note that children are *not* copied; they're pulled across when
445
def __init__(self, file_id, name, parent_id):
446
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
448
self.kind = 'directory'
450
def kind_character(self):
451
"""See InventoryEntry.kind_character."""
454
def _put_in_tar(self, item, tree):
455
"""See InventoryEntry._put_in_tar."""
456
item.type = tarfile.DIRTYPE
463
def _put_on_disk(self, fullpath, tree):
464
"""See InventoryEntry._put_on_disk."""
468
class InventoryFile(InventoryEntry):
469
"""A file in an inventory."""
471
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
472
'text_id', 'parent_id', 'children', 'executable',
473
'revision', 'symlink_target', 'reference_revision']
475
def _check(self, checker, tree_revision_id, tree):
476
"""See InventoryEntry._check"""
477
key = (self.file_id, self.revision)
478
if key in checker.checked_texts:
479
prev_sha = checker.checked_texts[key]
480
if prev_sha != self.text_sha1:
482
'mismatched sha1 on {%s} in {%s} (%s != %s) %r' %
483
(self.file_id, tree_revision_id, prev_sha, self.text_sha1,
486
checker.repeated_text_cnt += 1
489
mutter('check version {%s} of {%s}', tree_revision_id, self.file_id)
490
checker.checked_text_cnt += 1
491
# We can't check the length, because Weave doesn't store that
492
# information, and the whole point of looking at the weave's
493
# sha1sum is that we don't have to extract the text.
494
if (self.text_sha1 != tree._repository.texts.get_sha1s([key])[key]):
495
raise BzrCheckError('text {%s} version {%s} wrong sha1' % key)
496
checker.checked_texts[key] = self.text_sha1
499
other = InventoryFile(self.file_id, self.name, self.parent_id)
500
other.executable = self.executable
501
other.text_id = self.text_id
502
other.text_sha1 = self.text_sha1
503
other.text_size = self.text_size
504
other.revision = self.revision
507
def detect_changes(self, old_entry):
508
"""See InventoryEntry.detect_changes."""
509
text_modified = (self.text_sha1 != old_entry.text_sha1)
510
meta_modified = (self.executable != old_entry.executable)
511
return text_modified, meta_modified
513
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
514
output_to, reverse=False):
515
"""See InventoryEntry._diff."""
516
from bzrlib.diff import DiffText
517
from_file_id = self.file_id
519
to_file_id = to_entry.file_id
523
to_file_id, from_file_id = from_file_id, to_file_id
524
tree, to_tree = to_tree, tree
525
from_label, to_label = to_label, from_label
526
differ = DiffText(tree, to_tree, output_to, 'utf-8', '', '',
528
return differ.diff_text(from_file_id, to_file_id, from_label, to_label)
531
"""See InventoryEntry.has_text."""
534
def __init__(self, file_id, name, parent_id):
535
super(InventoryFile, self).__init__(file_id, name, parent_id)
538
def kind_character(self):
539
"""See InventoryEntry.kind_character."""
542
def _put_in_tar(self, item, tree):
543
"""See InventoryEntry._put_in_tar."""
544
item.type = tarfile.REGTYPE
545
fileobj = tree.get_file(self.file_id)
546
item.size = self.text_size
547
if tree.is_executable(self.file_id):
553
def _put_on_disk(self, fullpath, tree):
554
"""See InventoryEntry._put_on_disk."""
555
osutils.pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
556
if tree.is_executable(self.file_id):
557
os.chmod(fullpath, 0755)
559
def _read_tree_state(self, path, work_tree):
560
"""See InventoryEntry._read_tree_state."""
561
self.text_sha1 = work_tree.get_file_sha1(self.file_id, path=path)
562
# FIXME: 20050930 probe for the text size when getting sha1
563
# in _read_tree_state
564
self.executable = work_tree.is_executable(self.file_id, path=path)
567
return ("%s(%r, %r, parent_id=%r, sha1=%r, len=%s)"
568
% (self.__class__.__name__,
575
def _forget_tree_state(self):
576
self.text_sha1 = None
578
def _unchanged(self, previous_ie):
579
"""See InventoryEntry._unchanged."""
580
compatible = super(InventoryFile, self)._unchanged(previous_ie)
581
if self.text_sha1 != previous_ie.text_sha1:
584
# FIXME: 20050930 probe for the text size when getting sha1
585
# in _read_tree_state
586
self.text_size = previous_ie.text_size
587
if self.executable != previous_ie.executable:
592
class InventoryLink(InventoryEntry):
593
"""A file in an inventory."""
595
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
596
'text_id', 'parent_id', 'children', 'executable',
597
'revision', 'symlink_target', 'reference_revision']
599
def _check(self, checker, rev_id, tree):
600
"""See InventoryEntry._check"""
601
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
602
raise BzrCheckError('symlink {%s} has text in revision {%s}'
603
% (self.file_id, rev_id))
604
if self.symlink_target is None:
605
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
606
% (self.file_id, rev_id))
609
other = InventoryLink(self.file_id, self.name, self.parent_id)
610
other.symlink_target = self.symlink_target
611
other.revision = self.revision
614
def detect_changes(self, old_entry):
615
"""See InventoryEntry.detect_changes."""
616
# FIXME: which _modified field should we use ? RBC 20051003
617
text_modified = (self.symlink_target != old_entry.symlink_target)
619
mutter(" symlink target changed")
620
meta_modified = False
621
return text_modified, meta_modified
623
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
624
output_to, reverse=False):
625
"""See InventoryEntry._diff."""
626
from bzrlib.diff import DiffSymlink
627
old_target = self.symlink_target
628
if to_entry is not None:
629
new_target = to_entry.symlink_target
638
new_target, old_target = old_target, new_target
639
differ = DiffSymlink(old_tree, new_tree, output_to)
640
return differ.diff_symlink(old_target, new_target)
642
def __init__(self, file_id, name, parent_id):
643
super(InventoryLink, self).__init__(file_id, name, parent_id)
644
self.kind = 'symlink'
646
def kind_character(self):
647
"""See InventoryEntry.kind_character."""
650
def _put_in_tar(self, item, tree):
651
"""See InventoryEntry._put_in_tar."""
652
item.type = tarfile.SYMTYPE
656
item.linkname = self.symlink_target
659
def _put_on_disk(self, fullpath, tree):
660
"""See InventoryEntry._put_on_disk."""
662
os.symlink(self.symlink_target, fullpath)
664
raise BzrError("Failed to create symlink %r -> %r, error: %s" % (fullpath, self.symlink_target, e))
666
def _read_tree_state(self, path, work_tree):
667
"""See InventoryEntry._read_tree_state."""
668
self.symlink_target = work_tree.get_symlink_target(self.file_id)
670
def _forget_tree_state(self):
671
self.symlink_target = None
673
def _unchanged(self, previous_ie):
674
"""See InventoryEntry._unchanged."""
675
compatible = super(InventoryLink, self)._unchanged(previous_ie)
676
if self.symlink_target != previous_ie.symlink_target:
681
class TreeReference(InventoryEntry):
683
kind = 'tree-reference'
685
def __init__(self, file_id, name, parent_id, revision=None,
686
reference_revision=None):
687
InventoryEntry.__init__(self, file_id, name, parent_id)
688
self.revision = revision
689
self.reference_revision = reference_revision
692
return TreeReference(self.file_id, self.name, self.parent_id,
693
self.revision, self.reference_revision)
695
def _read_tree_state(self, path, work_tree):
696
"""Populate fields in the inventory entry from the given tree.
698
self.reference_revision = work_tree.get_reference_revision(
701
def _forget_tree_state(self):
702
self.reference_revision = None
704
def _unchanged(self, previous_ie):
705
"""See InventoryEntry._unchanged."""
706
compatible = super(TreeReference, self)._unchanged(previous_ie)
707
if self.reference_revision != previous_ie.reference_revision:
712
class Inventory(object):
713
"""Inventory of versioned files in a tree.
715
This describes which file_id is present at each point in the tree,
716
and possibly the SHA-1 or other information about the file.
717
Entries can be looked up either by path or by file_id.
719
The inventory represents a typical unix file tree, with
720
directories containing files and subdirectories. We never store
721
the full path to a file, because renaming a directory implicitly
722
moves all of its contents. This class internally maintains a
723
lookup tree that allows the children under a directory to be
726
InventoryEntry objects must not be modified after they are
727
inserted, other than through the Inventory API.
729
>>> inv = Inventory()
730
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
731
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
732
>>> inv['123-123'].name
735
May be treated as an iterator or set to look up file ids:
737
>>> bool(inv.path2id('hello.c'))
742
May also look up by name:
744
>>> [x[0] for x in inv.iter_entries()]
746
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
747
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
748
Traceback (most recent call last):
749
BzrError: parent_id {TREE_ROOT} not in inventory
750
>>> inv.add(InventoryFile('123-123', 'hello.c', 'TREE_ROOT-12345678-12345678'))
751
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT-12345678-12345678', sha1=None, len=None)
753
def __init__(self, root_id=ROOT_ID, revision_id=None):
754
"""Create or read an inventory.
756
If a working directory is specified, the inventory is read
757
from there. If the file is specified, read from that. If not,
758
the inventory is created empty.
760
The inventory is created with a default root directory, with
763
if root_id is not None:
764
self._set_root(InventoryDirectory(root_id, u'', None))
768
self.revision_id = revision_id
771
return "<Inventory object at %x, contents=%r>" % (id(self), self._byid)
773
def apply_delta(self, delta):
774
"""Apply a delta to this inventory.
776
:param delta: A list of changes to apply. After all the changes are
777
applied the final inventory must be internally consistent, but it
778
is ok to supply changes which, if only half-applied would have an
779
invalid result - such as supplying two changes which rename two
780
files, 'A' and 'B' with each other : [('A', 'B', 'A-id', a_entry),
781
('B', 'A', 'B-id', b_entry)].
783
Each change is a tuple, of the form (old_path, new_path, file_id,
786
When new_path is None, the change indicates the removal of an entry
787
from the inventory and new_entry will be ignored (using None is
788
appropriate). If new_path is not None, then new_entry must be an
789
InventoryEntry instance, which will be incorporated into the
790
inventory (and replace any existing entry with the same file id).
792
When old_path is None, the change indicates the addition of
793
a new entry to the inventory.
795
When neither new_path nor old_path are None, the change is a
796
modification to an entry, such as a rename, reparent, kind change
799
The children attribute of new_entry is ignored. This is because
800
this method preserves children automatically across alterations to
801
the parent of the children, and cases where the parent id of a
802
child is changing require the child to be passed in as a separate
803
change regardless. E.g. in the recursive deletion of a directory -
804
the directory's children must be included in the delta, or the
805
final inventory will be invalid.
808
# Remove all affected items which were in the original inventory,
809
# starting with the longest paths, thus ensuring parents are examined
810
# after their children, which means that everything we examine has no
811
# modified children remaining by the time we examine it.
812
for old_path, file_id in sorted(((op, f) for op, np, f, e in delta
813
if op is not None), reverse=True):
814
if file_id not in self:
817
# Preserve unaltered children of file_id for later reinsertion.
818
children[file_id] = getattr(self[file_id], 'children', {})
819
# Remove file_id and the unaltered children. If file_id is not
820
# being deleted it will be reinserted back later.
821
self.remove_recursive_id(file_id)
822
# Insert all affected which should be in the new inventory, reattaching
823
# their children if they had any. This is done from shortest path to
824
# longest, ensuring that items which were modified and whose parents in
825
# the resulting inventory were also modified, are inserted after their
827
for new_path, new_entry in sorted((np, e) for op, np, f, e in
828
delta if np is not None):
829
if new_entry.kind == 'directory':
830
new_entry.children = children.get(new_entry.file_id, {})
833
def _set_root(self, ie):
835
self._byid = {self.root.file_id: self.root}
838
# TODO: jam 20051218 Should copy also copy the revision_id?
839
entries = self.iter_entries()
840
if self.root is None:
841
return Inventory(root_id=None)
842
other = Inventory(entries.next()[1].file_id)
843
# copy recursively so we know directories will be added before
844
# their children. There are more efficient ways than this...
845
for path, entry in entries:
846
other.add(entry.copy())
850
return iter(self._byid)
853
"""Returns number of entries."""
854
return len(self._byid)
856
def iter_entries(self, from_dir=None):
857
"""Return (path, entry) pairs, in order by name."""
859
if self.root is None:
863
elif isinstance(from_dir, basestring):
864
from_dir = self._byid[from_dir]
866
# unrolling the recursive called changed the time from
867
# 440ms/663ms (inline/total) to 116ms/116ms
868
children = from_dir.children.items()
870
children = collections.deque(children)
871
stack = [(u'', children)]
873
from_dir_relpath, children = stack[-1]
876
name, ie = children.popleft()
878
# we know that from_dir_relpath never ends in a slash
879
# and 'f' doesn't begin with one, we can do a string op, rather
880
# than the checks of pathjoin(), though this means that all paths
882
path = from_dir_relpath + '/' + name
886
if ie.kind != 'directory':
889
# But do this child first
890
new_children = ie.children.items()
892
new_children = collections.deque(new_children)
893
stack.append((path, new_children))
894
# Break out of inner loop, so that we start outer loop with child
897
# if we finished all children, pop it off the stack
900
def iter_entries_by_dir(self, from_dir=None, specific_file_ids=None,
901
yield_parents=False):
902
"""Iterate over the entries in a directory first order.
904
This returns all entries for a directory before returning
905
the entries for children of a directory. This is not
906
lexicographically sorted order, and is a hybrid between
907
depth-first and breadth-first.
909
:param yield_parents: If True, yield the parents from the root leading
910
down to specific_file_ids that have been requested. This has no
911
impact if specific_file_ids is None.
912
:return: This yields (path, entry) pairs
914
if specific_file_ids and not isinstance(specific_file_ids, set):
915
specific_file_ids = set(specific_file_ids)
916
# TODO? Perhaps this should return the from_dir so that the root is
917
# yielded? or maybe an option?
919
if self.root is None:
921
# Optimize a common case
922
if (not yield_parents and specific_file_ids is not None and
923
len(specific_file_ids) == 1):
924
file_id = list(specific_file_ids)[0]
926
yield self.id2path(file_id), self[file_id]
929
if (specific_file_ids is None or yield_parents or
930
self.root.file_id in specific_file_ids):
932
elif isinstance(from_dir, basestring):
933
from_dir = self._byid[from_dir]
935
if specific_file_ids is not None:
936
# TODO: jam 20070302 This could really be done as a loop rather
937
# than a bunch of recursive calls.
940
def add_ancestors(file_id):
941
if file_id not in byid:
943
parent_id = byid[file_id].parent_id
944
if parent_id is None:
946
if parent_id not in parents:
947
parents.add(parent_id)
948
add_ancestors(parent_id)
949
for file_id in specific_file_ids:
950
add_ancestors(file_id)
954
stack = [(u'', from_dir)]
956
cur_relpath, cur_dir = stack.pop()
959
for child_name, child_ie in sorted(cur_dir.children.iteritems()):
961
child_relpath = cur_relpath + child_name
963
if (specific_file_ids is None or
964
child_ie.file_id in specific_file_ids or
965
(yield_parents and child_ie.file_id in parents)):
966
yield child_relpath, child_ie
968
if child_ie.kind == 'directory':
969
if parents is None or child_ie.file_id in parents:
970
child_dirs.append((child_relpath+'/', child_ie))
971
stack.extend(reversed(child_dirs))
973
def make_entry(self, kind, name, parent_id, file_id=None):
974
"""Simple thunk to bzrlib.inventory.make_entry."""
975
return make_entry(kind, name, parent_id, file_id)
978
"""Return list of (path, ie) for all entries except the root.
980
This may be faster than iter_entries.
983
def descend(dir_ie, dir_path):
984
kids = dir_ie.children.items()
986
for name, ie in kids:
987
child_path = osutils.pathjoin(dir_path, name)
988
accum.append((child_path, ie))
989
if ie.kind == 'directory':
990
descend(ie, child_path)
992
descend(self.root, u'')
995
def directories(self):
996
"""Return (path, entry) pairs for all directories, including the root.
999
def descend(parent_ie, parent_path):
1000
accum.append((parent_path, parent_ie))
1002
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
1005
for name, child_ie in kids:
1006
child_path = osutils.pathjoin(parent_path, name)
1007
descend(child_ie, child_path)
1008
descend(self.root, u'')
1011
def __contains__(self, file_id):
1012
"""True if this entry contains a file with given id.
1014
>>> inv = Inventory()
1015
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1016
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
1022
return (file_id in self._byid)
1024
def __getitem__(self, file_id):
1025
"""Return the entry for given file_id.
1027
>>> inv = Inventory()
1028
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
1029
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
1030
>>> inv['123123'].name
1034
return self._byid[file_id]
1036
# really we're passing an inventory, not a tree...
1037
raise errors.NoSuchId(self, file_id)
1039
def get_file_kind(self, file_id):
1040
return self._byid[file_id].kind
1042
def get_child(self, parent_id, filename):
1043
return self[parent_id].children.get(filename)
1045
def _add_child(self, entry):
1046
"""Add an entry to the inventory, without adding it to its parent"""
1047
if entry.file_id in self._byid:
1048
raise BzrError("inventory already contains entry with id {%s}" %
1050
self._byid[entry.file_id] = entry
1051
for child in getattr(entry, 'children', {}).itervalues():
1052
self._add_child(child)
1055
def add(self, entry):
1056
"""Add entry to inventory.
1058
To add a file to a branch ready to be committed, use Branch.add,
1061
Returns the new entry object.
1063
if entry.file_id in self._byid:
1064
raise errors.DuplicateFileId(entry.file_id,
1065
self._byid[entry.file_id])
1067
if entry.parent_id is None:
1071
parent = self._byid[entry.parent_id]
1073
raise BzrError("parent_id {%s} not in inventory" %
1076
if entry.name in parent.children:
1077
raise BzrError("%s is already versioned" %
1078
osutils.pathjoin(self.id2path(parent.file_id),
1079
entry.name).encode('utf-8'))
1080
parent.children[entry.name] = entry
1081
return self._add_child(entry)
1083
def add_path(self, relpath, kind, file_id=None, parent_id=None):
1084
"""Add entry from a path.
1086
The immediate parent must already be versioned.
1088
Returns the new entry object."""
1090
parts = osutils.splitpath(relpath)
1094
file_id = generate_ids.gen_root_id()
1095
self.root = InventoryDirectory(file_id, '', None)
1096
self._byid = {self.root.file_id: self.root}
1099
parent_path = parts[:-1]
1100
parent_id = self.path2id(parent_path)
1101
if parent_id is None:
1102
raise errors.NotVersionedError(path=parent_path)
1103
ie = make_entry(kind, parts[-1], parent_id, file_id)
1106
def __delitem__(self, file_id):
1107
"""Remove entry by id.
1109
>>> inv = Inventory()
1110
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1111
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
1119
del self._byid[file_id]
1120
if ie.parent_id is not None:
1121
del self[ie.parent_id].children[ie.name]
1123
def __eq__(self, other):
1124
"""Compare two sets by comparing their contents.
1126
>>> i1 = Inventory()
1127
>>> i2 = Inventory()
1130
>>> i1.add(InventoryFile('123', 'foo', ROOT_ID))
1131
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
1134
>>> i2.add(InventoryFile('123', 'foo', ROOT_ID))
1135
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
1139
if not isinstance(other, Inventory):
1140
return NotImplemented
1142
return self._byid == other._byid
1144
def __ne__(self, other):
1145
return not self.__eq__(other)
1148
raise ValueError('not hashable')
1150
def _iter_file_id_parents(self, file_id):
1151
"""Yield the parents of file_id up to the root."""
1152
while file_id is not None:
1154
ie = self._byid[file_id]
1156
raise errors.NoSuchId(tree=None, file_id=file_id)
1158
file_id = ie.parent_id
1160
def get_idpath(self, file_id):
1161
"""Return a list of file_ids for the path to an entry.
1163
The list contains one element for each directory followed by
1164
the id of the file itself. So the length of the returned list
1165
is equal to the depth of the file in the tree, counting the
1166
root directory as depth 1.
1169
for parent in self._iter_file_id_parents(file_id):
1170
p.insert(0, parent.file_id)
1173
def id2path(self, file_id):
1174
"""Return as a string the path to file_id.
1177
>>> e = i.add(InventoryDirectory('src-id', 'src', ROOT_ID))
1178
>>> e = i.add(InventoryFile('foo-id', 'foo.c', parent_id='src-id'))
1179
>>> print i.id2path('foo-id')
1182
# get all names, skipping root
1183
return '/'.join(reversed(
1184
[parent.name for parent in
1185
self._iter_file_id_parents(file_id)][:-1]))
1187
def path2id(self, name):
1188
"""Walk down through directories to return entry of last component.
1190
names may be either a list of path components, or a single
1191
string, in which case it is automatically split.
1193
This returns the entry of the last component in the path,
1194
which may be either a file or a directory.
1196
Returns None IFF the path is not found.
1198
if isinstance(name, basestring):
1199
name = osutils.splitpath(name)
1201
# mutter("lookup path %r" % name)
1208
children = getattr(parent, 'children', None)
1209
if children is None:
1214
# or raise an error?
1217
return parent.file_id
1219
def has_filename(self, names):
1220
return bool(self.path2id(names))
1222
def has_id(self, file_id):
1223
return (file_id in self._byid)
1225
def remove_recursive_id(self, file_id):
1226
"""Remove file_id, and children, from the inventory.
1228
:param file_id: A file_id to remove.
1230
to_find_delete = [self._byid[file_id]]
1232
while to_find_delete:
1233
ie = to_find_delete.pop()
1234
to_delete.append(ie.file_id)
1235
if ie.kind == 'directory':
1236
to_find_delete.extend(ie.children.values())
1237
for file_id in reversed(to_delete):
1239
del self._byid[file_id]
1240
if ie.parent_id is not None:
1241
del self[ie.parent_id].children[ie.name]
1245
def rename(self, file_id, new_parent_id, new_name):
1246
"""Move a file within the inventory.
1248
This can change either the name, or the parent, or both.
1250
This does not move the working file.
1252
new_name = ensure_normalized_name(new_name)
1253
if not is_valid_name(new_name):
1254
raise BzrError("not an acceptable filename: %r" % new_name)
1256
new_parent = self._byid[new_parent_id]
1257
if new_name in new_parent.children:
1258
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
1260
new_parent_idpath = self.get_idpath(new_parent_id)
1261
if file_id in new_parent_idpath:
1262
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
1263
% (self.id2path(file_id), self.id2path(new_parent_id)))
1265
file_ie = self._byid[file_id]
1266
old_parent = self._byid[file_ie.parent_id]
1268
# TODO: Don't leave things messed up if this fails
1270
del old_parent.children[file_ie.name]
1271
new_parent.children[new_name] = file_ie
1273
file_ie.name = new_name
1274
file_ie.parent_id = new_parent_id
1276
def is_root(self, file_id):
1277
return self.root is not None and file_id == self.root.file_id
1281
'directory': InventoryDirectory,
1282
'file': InventoryFile,
1283
'symlink': InventoryLink,
1284
'tree-reference': TreeReference
1287
def make_entry(kind, name, parent_id, file_id=None):
1288
"""Create an inventory entry.
1290
:param kind: the type of inventory entry to create.
1291
:param name: the basename of the entry.
1292
:param parent_id: the parent_id of the entry.
1293
:param file_id: the file_id to use. if None, one will be created.
1296
file_id = generate_ids.gen_file_id(name)
1297
name = ensure_normalized_name(name)
1299
factory = entry_factory[kind]
1301
raise BzrError("unknown kind %r" % kind)
1302
return factory(file_id, name, parent_id)
1305
def ensure_normalized_name(name):
1308
:raises InvalidNormalization: When name is not normalized, and cannot be
1309
accessed on this platform by the normalized path.
1310
:return: The NFC normalised version of name.
1312
#------- This has been copied to bzrlib.dirstate.DirState.add, please
1313
# keep them synchronised.
1314
# we dont import normalized_filename directly because we want to be
1315
# able to change the implementation at runtime for tests.
1316
norm_name, can_access = osutils.normalized_filename(name)
1317
if norm_name != name:
1321
# TODO: jam 20060701 This would probably be more useful
1322
# if the error was raised with the full path
1323
raise errors.InvalidNormalization(name)
1329
def is_valid_name(name):
1331
if _NAME_RE is None:
1332
_NAME_RE = re.compile(r'^[^/\\]+$')
1334
return bool(_NAME_RE.match(name))
added
removed
bzrlib/inventory.py
