36
60
An InventoryEntry has the following fields, which are also
37
61
present in the XML inventory-entry element:
40
* *name*: (only the basename within the directory, must not
42
* *kind*: "directory" or "file"
43
* *directory_id*: (if absent/null means the branch root directory)
44
* *text_sha1*: only for files
45
* *text_size*: in bytes, only for files
46
* *text_id*: identifier for the text version, only for files
48
InventoryEntries can also exist inside a WorkingTree
49
inventory, in which case they are not yet bound to a
50
particular revision of the file. In that case the text_sha1,
51
text_size and text_id are absent.
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.)
54
87
>>> i = Inventory()
57
>>> i.add(InventoryEntry('123', 'src', 'directory', ROOT_ID))
58
>>> i.add(InventoryEntry('2323', 'hello.c', 'file', parent_id='123'))
59
>>> for j in i.iter_entries():
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])
62
('src', InventoryEntry('123', 'src', kind='directory', parent_id='TREE_ROOT'))
63
('src/hello.c', InventoryEntry('2323', 'hello.c', kind='file', parent_id='123'))
64
>>> i.add(InventoryEntry('2323', 'bye.c', 'file', '123'))
65
Traceback (most recent call last):
67
BzrError: inventory already contains entry with id {2323}
68
>>> i.add(InventoryEntry('2324', 'bye.c', 'file', '123'))
69
>>> i.add(InventoryEntry('2325', 'wibble', 'directory', '123'))
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)
70
105
>>> i.path2id('src/wibble')
74
>>> i.add(InventoryEntry('2326', 'wibble.c', 'file', '2325'))
109
>>> i.add(InventoryFile('2326', 'wibble.c', '2325'))
110
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
76
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
77
>>> for j in i.iter_entries():
79
... assert i.path2id(j[0])
112
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
113
>>> for path, entry in i.iter_entries():
115
... assert i.path2id(path)
85
122
src/wibble/wibble.c
86
123
>>> i.id2path('2326')
87
124
'src/wibble/wibble.c'
89
TODO: Maybe also keep the full path of the entry, and the children?
90
But those depend on its position within a particular inventory, and
91
it would be nice not to need to hold the backpointer here.
94
# TODO: split InventoryEntry into subclasses for files,
95
# directories, etc etc.
97
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
98
'text_id', 'parent_id', 'children', ]
100
def __init__(self, file_id, name, kind, parent_id, text_id=None):
127
# Constants returned by describe_change()
129
# TODO: These should probably move to some kind of FileChangeDescription
130
# class; that's like what's inside a TreeDelta but we want to be able to
131
# generate them just for one file at a time.
133
MODIFIED_AND_RENAMED = 'modified and renamed'
137
def detect_changes(self, old_entry):
138
"""Return a (text_modified, meta_modified) from this to old_entry.
140
_read_tree_state must have been called on self and old_entry prior to
141
calling detect_changes.
145
def diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
146
output_to, reverse=False):
147
"""Perform a diff from this to to_entry.
149
text_diff will be used for textual difference calculation.
150
This is a template method, override _diff in child classes.
152
self._read_tree_state(tree.id2path(self.file_id), tree)
154
# cannot diff from one kind to another - you must do a removal
155
# and an addif they do not match.
156
assert self.kind == to_entry.kind
157
to_entry._read_tree_state(to_tree.id2path(to_entry.file_id),
159
self._diff(text_diff, from_label, tree, to_label, to_entry, to_tree,
162
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
163
output_to, reverse=False):
164
"""Perform a diff between two entries of the same kind."""
166
def parent_candidates(self, previous_inventories):
167
"""Find possible per-file graph parents.
169
This is currently defined by:
170
- Select the last changed revision in the parent inventory.
171
- Do deal with a short lived bug in bzr 0.8's development two entries
172
that have the same last changed but different 'x' bit settings are
175
# revision:ie mapping for each ie found in previous_inventories.
177
# identify candidate head revision ids.
178
for inv in previous_inventories:
179
if self.file_id in inv:
180
ie = inv[self.file_id]
181
assert ie.file_id == self.file_id
182
if ie.revision in candidates:
183
# same revision value in two different inventories:
184
# correct possible inconsistencies:
185
# * there was a bug in revision updates with 'x' bit
188
if candidates[ie.revision].executable != ie.executable:
189
candidates[ie.revision].executable = False
190
ie.executable = False
191
except AttributeError:
193
# must now be the same.
194
assert candidates[ie.revision] == ie
196
# add this revision as a candidate.
197
candidates[ie.revision] = ie
200
@deprecated_method(zero_ninetyone)
201
def find_previous_heads(self, previous_inventories,
202
versioned_file_store,
205
"""Return the revisions and entries that directly precede this.
207
Returned as a map from revision to inventory entry.
209
This is a map containing the file revisions in all parents
210
for which the file exists, and its revision is not a parent of
211
any other. If the file is new, the set will be empty.
213
:param versioned_file_store: A store where ancestry data on this
214
file id can be queried.
215
:param transaction: The transaction that queries to the versioned
216
file store should be completed under.
217
:param entry_vf: The entry versioned file, if its already available.
219
candidates = self.parent_candidates(previous_inventories)
221
# revision:ie mapping with one revision for each head.
223
# common case optimisation
224
if len(candidates) == 1:
225
# if there is only one candidate revision found
226
# then we can avoid opening the versioned file to access ancestry:
227
# there cannot be any ancestors to eliminate when there is
228
# only one revision available.
231
# --- what follows is now encapsulated in repository.get_graph.heads(),
232
# but that is not accessible from here as we have no repository
233
# pointer. Note that the repository.get_graph.heads() call can return
234
# different results *at the moment* because of the kind-changing check
235
# we have in parent_candidates().
237
# eliminate ancestors amongst the available candidates:
238
# heads are those that are not an ancestor of any other candidate
239
# - this provides convergence at a per-file level.
240
def get_ancestors(weave, entry):
241
return set(weave.get_ancestry(entry.revision, topo_sorted=False))
242
# revision: ancestor list for each head
244
for ie in candidates.values():
245
# may be an ancestor of a known head:
246
already_present = 0 != len(
247
[head for head in heads
248
if ie.revision in head_ancestors[head]])
250
# an ancestor of an analyzed candidate.
252
# not an ancestor of a known head:
253
# load the versioned file for this file id if needed
255
entry_vf = versioned_file_store.get_weave_or_empty(
256
self.file_id, transaction)
257
ancestors = get_ancestors(entry_vf, ie)
258
# may knock something else out:
259
check_heads = list(heads.keys())
260
for head in check_heads:
261
if head in ancestors:
262
# this previously discovered 'head' is not
263
# really a head - its an ancestor of the newly
266
head_ancestors[ie.revision] = ancestors
267
heads[ie.revision] = ie
270
def get_tar_item(self, root, dp, now, tree):
271
"""Get a tarfile item and a file stream for its content."""
272
item = tarfile.TarInfo(osutils.pathjoin(root, dp).encode('utf8'))
273
# TODO: would be cool to actually set it to the timestamp of the
274
# revision it was last changed
276
fileobj = self._put_in_tar(item, tree)
280
"""Return true if the object this entry represents has textual data.
282
Note that textual data includes binary content.
284
Also note that all entries get weave files created for them.
285
This attribute is primarily used when upgrading from old trees that
286
did not have the weave index for all inventory entries.
290
def __init__(self, file_id, name, parent_id, text_id=None):
101
291
"""Create an InventoryEntry
103
293
The filename must be a single component, relative to the
104
294
parent directory; it cannot be a whole path or relative name.
106
>>> e = InventoryEntry('123', 'hello.c', 'file', ROOT_ID)
296
>>> e = InventoryFile('123', 'hello.c', ROOT_ID)
111
>>> e = InventoryEntry('123', 'src/hello.c', 'file', ROOT_ID)
301
>>> e = InventoryFile('123', 'src/hello.c', ROOT_ID)
112
302
Traceback (most recent call last):
113
BzrCheckError: InventoryEntry name 'src/hello.c' is invalid
303
InvalidEntryName: Invalid entry name: src/hello.c
305
assert isinstance(name, basestring), name
115
306
if '/' in name or '\\' in name:
116
raise BzrCheckError('InventoryEntry name %r is invalid' % name)
307
raise errors.InvalidEntryName(name=name)
308
self.executable = False
118
310
self.text_sha1 = None
119
311
self.text_size = None
121
312
self.file_id = file_id
313
assert isinstance(file_id, (str, None.__class__)), \
314
'bad type %r for %r' % (type(file_id), file_id)
124
316
self.text_id = text_id
125
317
self.parent_id = parent_id
126
if kind == 'directory':
131
raise BzrError("unhandled entry kind %r" % kind)
318
self.symlink_target = None
319
self.reference_revision = None
321
def kind_character(self):
322
"""Return a short kind indicator useful for appending to names."""
323
raise BzrError('unknown kind %r' % self.kind)
325
known_kinds = ('file', 'directory', 'symlink')
327
def _put_in_tar(self, item, tree):
328
"""populate item for stashing in a tar, and return the content stream.
330
If no content is available, return None.
332
raise BzrError("don't know how to export {%s} of kind %r" %
333
(self.file_id, self.kind))
335
def put_on_disk(self, dest, dp, tree):
336
"""Create a representation of self on disk in the prefix dest.
338
This is a template method - implement _put_on_disk in subclasses.
340
fullpath = osutils.pathjoin(dest, dp)
341
self._put_on_disk(fullpath, tree)
342
# mutter(" export {%s} kind %s to %s", self.file_id,
343
# self.kind, fullpath)
345
def _put_on_disk(self, fullpath, tree):
346
"""Put this entry onto disk at fullpath, from tree tree."""
347
raise BzrError("don't know how to export {%s} of kind %r" % (self.file_id, self.kind))
135
349
def sorted_children(self):
136
l = self.children.items()
350
return sorted(self.children.items())
353
def versionable_kind(kind):
354
return (kind in ('file', 'directory', 'symlink', 'tree-reference'))
356
def check(self, checker, rev_id, inv, tree):
357
"""Check this inventory entry is intact.
359
This is a template method, override _check for kind specific
362
:param checker: Check object providing context for the checks;
363
can be used to find out what parts of the repository have already
365
:param rev_id: Revision id from which this InventoryEntry was loaded.
366
Not necessarily the last-changed revision for this file.
367
:param inv: Inventory from which the entry was loaded.
368
:param tree: RevisionTree for this entry.
370
if self.parent_id is not None:
371
if not inv.has_id(self.parent_id):
372
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
373
% (self.parent_id, rev_id))
374
self._check(checker, rev_id, tree)
376
def _check(self, checker, rev_id, tree):
377
"""Check this inventory entry for kind specific errors."""
378
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
142
other = InventoryEntry(self.file_id, self.name, self.kind,
143
self.parent_id, text_id=self.text_id)
144
other.text_sha1 = self.text_sha1
145
other.text_size = self.text_size
146
# note that children are *not* copied; they're pulled across when
382
"""Clone this inventory entry."""
383
raise NotImplementedError
386
def describe_change(old_entry, new_entry):
387
"""Describe the change between old_entry and this.
389
This smells of being an InterInventoryEntry situation, but as its
390
the first one, we're making it a static method for now.
392
An entry with a different parent, or different name is considered
393
to be renamed. Reparenting is an internal detail.
394
Note that renaming the parent does not trigger a rename for the
397
# TODO: Perhaps return an object rather than just a string
398
if old_entry is new_entry:
399
# also the case of both being None
401
elif old_entry is None:
403
elif new_entry is None:
405
if old_entry.kind != new_entry.kind:
407
text_modified, meta_modified = new_entry.detect_changes(old_entry)
408
if text_modified or meta_modified:
412
# TODO 20060511 (mbp, rbc) factor out 'detect_rename' here.
413
if old_entry.parent_id != new_entry.parent_id:
415
elif old_entry.name != new_entry.name:
419
if renamed and not modified:
420
return InventoryEntry.RENAMED
421
if modified and not renamed:
423
if modified and renamed:
424
return InventoryEntry.MODIFIED_AND_RENAMED
151
427
def __repr__(self):
152
return ("%s(%r, %r, kind=%r, parent_id=%r)"
428
return ("%s(%r, %r, parent_id=%r, revision=%r)"
153
429
% (self.__class__.__name__,
160
def to_element(self):
161
"""Convert to XML element"""
162
from bzrlib.xml import Element
166
e.set('name', self.name)
167
e.set('file_id', self.file_id)
168
e.set('kind', self.kind)
170
if self.text_size != None:
171
e.set('text_size', '%d' % self.text_size)
173
for f in ['text_id', 'text_sha1']:
178
# to be conservative, we don't externalize the root pointers
179
# for now, leaving them as null in the xml form. in a future
180
# version it will be implied by nested elements.
181
if self.parent_id != ROOT_ID:
182
assert isinstance(self.parent_id, basestring)
183
e.set('parent_id', self.parent_id)
190
def from_element(cls, elt):
191
assert elt.tag == 'entry'
193
## original format inventories don't have a parent_id for
194
## nodes in the root directory, but it's cleaner to use one
196
parent_id = elt.get('parent_id')
197
if parent_id == None:
200
self = cls(elt.get('file_id'), elt.get('name'), elt.get('kind'), parent_id)
201
self.text_id = elt.get('text_id')
202
self.text_sha1 = elt.get('text_sha1')
204
## mutter("read inventoryentry: %r" % (elt.attrib))
206
v = elt.get('text_size')
207
self.text_size = v and int(v)
212
from_element = classmethod(from_element)
214
435
def __eq__(self, other):
215
436
if not isinstance(other, InventoryEntry):
216
437
return NotImplemented
218
return (self.file_id == other.file_id) \
219
and (self.name == other.name) \
220
and (self.text_sha1 == other.text_sha1) \
221
and (self.text_size == other.text_size) \
222
and (self.text_id == other.text_id) \
223
and (self.parent_id == other.parent_id) \
224
and (self.kind == other.kind)
439
return ((self.file_id == other.file_id)
440
and (self.name == other.name)
441
and (other.symlink_target == self.symlink_target)
442
and (self.text_sha1 == other.text_sha1)
443
and (self.text_size == other.text_size)
444
and (self.text_id == other.text_id)
445
and (self.parent_id == other.parent_id)
446
and (self.kind == other.kind)
447
and (self.revision == other.revision)
448
and (self.executable == other.executable)
449
and (self.reference_revision == other.reference_revision)
227
452
def __ne__(self, other):
228
453
return not (self == other)
248
514
and (self.children == other.children)
517
class InventoryDirectory(InventoryEntry):
518
"""A directory in an inventory."""
520
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
521
'text_id', 'parent_id', 'children', 'executable',
522
'revision', 'symlink_target', 'reference_revision']
524
def _check(self, checker, rev_id, tree):
525
"""See InventoryEntry._check"""
526
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
527
raise BzrCheckError('directory {%s} has text in revision {%s}'
528
% (self.file_id, rev_id))
531
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
532
other.revision = self.revision
533
# note that children are *not* copied; they're pulled across when
537
def __init__(self, file_id, name, parent_id):
538
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
540
self.kind = 'directory'
542
def kind_character(self):
543
"""See InventoryEntry.kind_character."""
546
def _put_in_tar(self, item, tree):
547
"""See InventoryEntry._put_in_tar."""
548
item.type = tarfile.DIRTYPE
555
def _put_on_disk(self, fullpath, tree):
556
"""See InventoryEntry._put_on_disk."""
560
class InventoryFile(InventoryEntry):
561
"""A file in an inventory."""
563
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
564
'text_id', 'parent_id', 'children', 'executable',
565
'revision', 'symlink_target', 'reference_revision']
567
def _check(self, checker, tree_revision_id, tree):
568
"""See InventoryEntry._check"""
569
t = (self.file_id, self.revision)
570
if t in checker.checked_texts:
571
prev_sha = checker.checked_texts[t]
572
if prev_sha != self.text_sha1:
574
'mismatched sha1 on {%s} in {%s} (%s != %s) %r' %
575
(self.file_id, tree_revision_id, prev_sha, self.text_sha1,
578
checker.repeated_text_cnt += 1
581
if self.file_id not in checker.checked_weaves:
582
mutter('check weave {%s}', self.file_id)
583
w = tree._get_weave(self.file_id)
584
# Not passing a progress bar, because it creates a new
585
# progress, which overwrites the current progress,
586
# and doesn't look nice
588
checker.checked_weaves[self.file_id] = True
590
w = tree._get_weave(self.file_id)
592
mutter('check version {%s} of {%s}', tree_revision_id, self.file_id)
593
checker.checked_text_cnt += 1
594
# We can't check the length, because Weave doesn't store that
595
# information, and the whole point of looking at the weave's
596
# sha1sum is that we don't have to extract the text.
597
if self.text_sha1 != w.get_sha1(self.revision):
598
raise BzrCheckError('text {%s} version {%s} wrong sha1'
599
% (self.file_id, self.revision))
600
checker.checked_texts[t] = self.text_sha1
603
other = InventoryFile(self.file_id, self.name, self.parent_id)
604
other.executable = self.executable
605
other.text_id = self.text_id
606
other.text_sha1 = self.text_sha1
607
other.text_size = self.text_size
608
other.revision = self.revision
611
def detect_changes(self, old_entry):
612
"""See InventoryEntry.detect_changes."""
613
assert self.text_sha1 is not None
614
assert old_entry.text_sha1 is not None
615
text_modified = (self.text_sha1 != old_entry.text_sha1)
616
meta_modified = (self.executable != old_entry.executable)
617
return text_modified, meta_modified
619
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
620
output_to, reverse=False):
621
"""See InventoryEntry._diff."""
622
from bzrlib.diff import DiffText
623
from_file_id = self.file_id
625
to_file_id = to_entry.file_id
629
to_file_id, from_file_id = from_file_id, to_file_id
630
tree, to_tree = to_tree, tree
631
from_label, to_label = to_label, from_label
632
differ = DiffText(tree, to_tree, output_to, 'utf-8', '', '',
634
return differ.diff_text(from_file_id, to_file_id, from_label, to_label)
637
"""See InventoryEntry.has_text."""
640
def __init__(self, file_id, name, parent_id):
641
super(InventoryFile, self).__init__(file_id, name, parent_id)
644
def kind_character(self):
645
"""See InventoryEntry.kind_character."""
648
def _put_in_tar(self, item, tree):
649
"""See InventoryEntry._put_in_tar."""
650
item.type = tarfile.REGTYPE
651
fileobj = tree.get_file(self.file_id)
652
item.size = self.text_size
653
if tree.is_executable(self.file_id):
659
def _put_on_disk(self, fullpath, tree):
660
"""See InventoryEntry._put_on_disk."""
661
osutils.pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
662
if tree.is_executable(self.file_id):
663
os.chmod(fullpath, 0755)
665
def _read_tree_state(self, path, work_tree):
666
"""See InventoryEntry._read_tree_state."""
667
self.text_sha1 = work_tree.get_file_sha1(self.file_id, path=path)
668
# FIXME: 20050930 probe for the text size when getting sha1
669
# in _read_tree_state
670
self.executable = work_tree.is_executable(self.file_id, path=path)
673
return ("%s(%r, %r, parent_id=%r, sha1=%r, len=%s)"
674
% (self.__class__.__name__,
681
def _forget_tree_state(self):
682
self.text_sha1 = None
684
def _unchanged(self, previous_ie):
685
"""See InventoryEntry._unchanged."""
686
compatible = super(InventoryFile, self)._unchanged(previous_ie)
687
if self.text_sha1 != previous_ie.text_sha1:
690
# FIXME: 20050930 probe for the text size when getting sha1
691
# in _read_tree_state
692
self.text_size = previous_ie.text_size
693
if self.executable != previous_ie.executable:
698
class InventoryLink(InventoryEntry):
699
"""A file in an inventory."""
701
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
702
'text_id', 'parent_id', 'children', 'executable',
703
'revision', 'symlink_target', 'reference_revision']
705
def _check(self, checker, rev_id, tree):
706
"""See InventoryEntry._check"""
707
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
708
raise BzrCheckError('symlink {%s} has text in revision {%s}'
709
% (self.file_id, rev_id))
710
if self.symlink_target is None:
711
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
712
% (self.file_id, rev_id))
715
other = InventoryLink(self.file_id, self.name, self.parent_id)
716
other.symlink_target = self.symlink_target
717
other.revision = self.revision
720
def detect_changes(self, old_entry):
721
"""See InventoryEntry.detect_changes."""
722
# FIXME: which _modified field should we use ? RBC 20051003
723
text_modified = (self.symlink_target != old_entry.symlink_target)
725
mutter(" symlink target changed")
726
meta_modified = False
727
return text_modified, meta_modified
729
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
730
output_to, reverse=False):
731
"""See InventoryEntry._diff."""
732
from bzrlib.diff import DiffSymlink
733
old_target = self.symlink_target
734
if to_entry is not None:
735
new_target = to_entry.symlink_target
744
new_target, old_target = old_target, new_target
745
differ = DiffSymlink(old_tree, new_tree, output_to)
746
return differ.diff_symlink(old_target, new_target)
748
def __init__(self, file_id, name, parent_id):
749
super(InventoryLink, self).__init__(file_id, name, parent_id)
750
self.kind = 'symlink'
752
def kind_character(self):
753
"""See InventoryEntry.kind_character."""
756
def _put_in_tar(self, item, tree):
757
"""See InventoryEntry._put_in_tar."""
758
item.type = tarfile.SYMTYPE
762
item.linkname = self.symlink_target
765
def _put_on_disk(self, fullpath, tree):
766
"""See InventoryEntry._put_on_disk."""
768
os.symlink(self.symlink_target, fullpath)
770
raise BzrError("Failed to create symlink %r -> %r, error: %s" % (fullpath, self.symlink_target, e))
772
def _read_tree_state(self, path, work_tree):
773
"""See InventoryEntry._read_tree_state."""
774
self.symlink_target = work_tree.get_symlink_target(self.file_id)
776
def _forget_tree_state(self):
777
self.symlink_target = None
779
def _unchanged(self, previous_ie):
780
"""See InventoryEntry._unchanged."""
781
compatible = super(InventoryLink, self)._unchanged(previous_ie)
782
if self.symlink_target != previous_ie.symlink_target:
787
class TreeReference(InventoryEntry):
789
kind = 'tree-reference'
791
def __init__(self, file_id, name, parent_id, revision=None,
792
reference_revision=None):
793
InventoryEntry.__init__(self, file_id, name, parent_id)
794
self.revision = revision
795
self.reference_revision = reference_revision
798
return TreeReference(self.file_id, self.name, self.parent_id,
799
self.revision, self.reference_revision)
801
def _read_tree_state(self, path, work_tree):
802
"""Populate fields in the inventory entry from the given tree.
804
self.reference_revision = work_tree.get_reference_revision(
807
def _forget_tree_state(self):
808
self.reference_revision = None
810
def _unchanged(self, previous_ie):
811
"""See InventoryEntry._unchanged."""
812
compatible = super(TreeReference, self)._unchanged(previous_ie)
813
if self.reference_revision != previous_ie.reference_revision:
252
818
class Inventory(object):
253
819
"""Inventory of versioned files in a tree.
295
866
The inventory is created with a default root directory, with
298
# We are letting Branch(init=True) create a unique inventory
299
# root id. Rather than generating a random one here.
301
# root_id = bzrlib.branch.gen_file_id('TREE_ROOT')
302
self.root = RootEntry(root_id)
869
if root_id is not None:
870
assert root_id.__class__ == str
871
self._set_root(InventoryDirectory(root_id, u'', None))
875
self.revision_id = revision_id
878
return "<Inventory object at %x, contents=%r>" % (id(self), self._byid)
880
def apply_delta(self, delta):
881
"""Apply a delta to this inventory.
883
:param delta: A list of changes to apply. After all the changes are
884
applied the final inventory must be internally consistent, but it
885
is ok to supply changes which, if only half-applied would have an
886
invalid result - such as supplying two changes which rename two
887
files, 'A' and 'B' with each other : [('A', 'B', 'A-id', a_entry),
888
('B', 'A', 'B-id', b_entry)].
890
Each change is a tuple, of the form (old_path, new_path, file_id,
893
When new_path is None, the change indicates the removal of an entry
894
from the inventory and new_entry will be ignored (using None is
895
appropriate). If new_path is not None, then new_entry must be an
896
InventoryEntry instance, which will be incorporated into the
897
inventory (and replace any existing entry with the same file id).
899
When old_path is None, the change indicates the addition of
900
a new entry to the inventory.
902
When neither new_path nor old_path are None, the change is a
903
modification to an entry, such as a rename, reparent, kind change
906
The children attribute of new_entry is ignored. This is because
907
this method preserves children automatically across alterations to
908
the parent of the children, and cases where the parent id of a
909
child is changing require the child to be passed in as a separate
910
change regardless. E.g. in the recursive deletion of a directory -
911
the directory's children must be included in the delta, or the
912
final inventory will be invalid.
915
# Remove all affected items which were in the original inventory,
916
# starting with the longest paths, thus ensuring parents are examined
917
# after their children, which means that everything we examine has no
918
# modified children remaining by the time we examine it.
919
for old_path, file_id in sorted(((op, f) for op, np, f, e in delta
920
if op is not None), reverse=True):
921
if file_id not in self:
924
# Preserve unaltered children of file_id for later reinsertion.
925
children[file_id] = getattr(self[file_id], 'children', {})
926
# Remove file_id and the unaltered children. If file_id is not
927
# being deleted it will be reinserted back later.
928
self.remove_recursive_id(file_id)
929
# Insert all affected which should be in the new inventory, reattaching
930
# their children if they had any. This is done from shortest path to
931
# longest, ensuring that items which were modified and whose parents in
932
# the resulting inventory were also modified, are inserted after their
934
for new_path, new_entry in sorted((np, e) for op, np, f, e in
935
delta if np is not None):
936
if new_entry.kind == 'directory':
937
new_entry.children = children.get(new_entry.file_id, {})
940
def _set_root(self, ie):
303
942
self._byid = {self.root.file_id: self.root}
945
# TODO: jam 20051218 Should copy also copy the revision_id?
946
entries = self.iter_entries()
947
if self.root is None:
948
return Inventory(root_id=None)
949
other = Inventory(entries.next()[1].file_id)
950
# copy recursively so we know directories will be added before
951
# their children. There are more efficient ways than this...
952
for path, entry in entries:
953
other.add(entry.copy())
306
956
def __iter__(self):
307
957
return iter(self._byid)
310
959
def __len__(self):
311
960
"""Returns number of entries."""
312
961
return len(self._byid)
315
963
def iter_entries(self, from_dir=None):
316
964
"""Return (path, entry) pairs, in order by name."""
320
elif isinstance(from_dir, basestring):
321
from_dir = self._byid[from_dir]
323
kids = from_dir.children.items()
325
for name, ie in kids:
327
if ie.kind == 'directory':
328
for cn, cie in self.iter_entries(from_dir=ie.file_id):
329
yield os.path.join(name, cn), cie
966
if self.root is None:
970
elif isinstance(from_dir, basestring):
971
from_dir = self._byid[from_dir]
973
# unrolling the recursive called changed the time from
974
# 440ms/663ms (inline/total) to 116ms/116ms
975
children = from_dir.children.items()
977
children = collections.deque(children)
978
stack = [(u'', children)]
980
from_dir_relpath, children = stack[-1]
983
name, ie = children.popleft()
985
# we know that from_dir_relpath never ends in a slash
986
# and 'f' doesn't begin with one, we can do a string op, rather
987
# than the checks of pathjoin(), though this means that all paths
989
path = from_dir_relpath + '/' + name
993
if ie.kind != 'directory':
996
# But do this child first
997
new_children = ie.children.items()
999
new_children = collections.deque(new_children)
1000
stack.append((path, new_children))
1001
# Break out of inner loop, so that we start outer loop with child
1004
# if we finished all children, pop it off the stack
1007
def iter_entries_by_dir(self, from_dir=None, specific_file_ids=None,
1008
yield_parents=False):
1009
"""Iterate over the entries in a directory first order.
1011
This returns all entries for a directory before returning
1012
the entries for children of a directory. This is not
1013
lexicographically sorted order, and is a hybrid between
1014
depth-first and breadth-first.
1016
:param yield_parents: If True, yield the parents from the root leading
1017
down to specific_file_ids that have been requested. This has no
1018
impact if specific_file_ids is None.
1019
:return: This yields (path, entry) pairs
1021
if specific_file_ids and not isinstance(specific_file_ids, set):
1022
specific_file_ids = set(specific_file_ids)
1023
# TODO? Perhaps this should return the from_dir so that the root is
1024
# yielded? or maybe an option?
1025
if from_dir is None:
1026
if self.root is None:
1028
# Optimize a common case
1029
if (not yield_parents and specific_file_ids is not None and
1030
len(specific_file_ids) == 1):
1031
file_id = list(specific_file_ids)[0]
1033
yield self.id2path(file_id), self[file_id]
1035
from_dir = self.root
1036
if (specific_file_ids is None or yield_parents or
1037
self.root.file_id in specific_file_ids):
1038
yield u'', self.root
1039
elif isinstance(from_dir, basestring):
1040
from_dir = self._byid[from_dir]
1042
if specific_file_ids is not None:
1043
# TODO: jam 20070302 This could really be done as a loop rather
1044
# than a bunch of recursive calls.
1047
def add_ancestors(file_id):
1048
if file_id not in byid:
1050
parent_id = byid[file_id].parent_id
1051
if parent_id is None:
1053
if parent_id not in parents:
1054
parents.add(parent_id)
1055
add_ancestors(parent_id)
1056
for file_id in specific_file_ids:
1057
add_ancestors(file_id)
1061
stack = [(u'', from_dir)]
1063
cur_relpath, cur_dir = stack.pop()
1066
for child_name, child_ie in sorted(cur_dir.children.iteritems()):
1068
child_relpath = cur_relpath + child_name
1070
if (specific_file_ids is None or
1071
child_ie.file_id in specific_file_ids or
1072
(yield_parents and child_ie.file_id in parents)):
1073
yield child_relpath, child_ie
1075
if child_ie.kind == 'directory':
1076
if parents is None or child_ie.file_id in parents:
1077
child_dirs.append((child_relpath+'/', child_ie))
1078
stack.extend(reversed(child_dirs))
1080
def make_entry(self, kind, name, parent_id, file_id=None):
1081
"""Simple thunk to bzrlib.inventory.make_entry."""
1082
return make_entry(kind, name, parent_id, file_id)
332
1084
def entries(self):
333
1085
"""Return list of (path, ie) for all entries except the root.
402
1149
def get_child(self, parent_id, filename):
403
1150
return self[parent_id].children.get(filename)
1152
def _add_child(self, entry):
1153
"""Add an entry to the inventory, without adding it to its parent"""
1154
if entry.file_id in self._byid:
1155
raise BzrError("inventory already contains entry with id {%s}" %
1157
self._byid[entry.file_id] = entry
1158
for child in getattr(entry, 'children', {}).itervalues():
1159
self._add_child(child)
406
1162
def add(self, entry):
407
1163
"""Add entry to inventory.
409
1165
To add a file to a branch ready to be committed, use Branch.add,
1168
Returns the new entry object.
411
1170
if entry.file_id in self._byid:
412
raise BzrError("inventory already contains entry with id {%s}" % entry.file_id)
414
if entry.parent_id == ROOT_ID or entry.parent_id is None:
415
entry.parent_id = self.root.file_id
418
parent = self._byid[entry.parent_id]
420
raise BzrError("parent_id {%s} not in inventory" % entry.parent_id)
422
if parent.children.has_key(entry.name):
423
raise BzrError("%s is already versioned" %
424
appendpath(self.id2path(parent.file_id), entry.name))
426
self._byid[entry.file_id] = entry
427
parent.children[entry.name] = entry
430
def add_path(self, relpath, kind, file_id=None):
1171
raise errors.DuplicateFileId(entry.file_id,
1172
self._byid[entry.file_id])
1174
if entry.parent_id is None:
1175
assert self.root is None and len(self._byid) == 0
1179
parent = self._byid[entry.parent_id]
1181
raise BzrError("parent_id {%s} not in inventory" %
1184
if entry.name in parent.children:
1185
raise BzrError("%s is already versioned" %
1186
osutils.pathjoin(self.id2path(parent.file_id),
1187
entry.name).encode('utf-8'))
1188
parent.children[entry.name] = entry
1189
return self._add_child(entry)
1191
def add_path(self, relpath, kind, file_id=None, parent_id=None):
431
1192
"""Add entry from a path.
433
The immediate parent must already be versioned"""
434
from bzrlib.branch import gen_file_id
1194
The immediate parent must already be versioned.
1196
Returns the new entry object."""
436
parts = bzrlib.osutils.splitpath(relpath)
1198
parts = osutils.splitpath(relpath)
437
1200
if len(parts) == 0:
438
raise BzrError("cannot re-add root of inventory")
441
file_id = gen_file_id(relpath)
443
parent_path = parts[:-1]
444
parent_id = self.path2id(parent_path)
445
if parent_id == None:
446
raise NotVersionedError(parent_path)
448
ie = InventoryEntry(file_id, parts[-1],
449
kind=kind, parent_id=parent_id)
1202
file_id = generate_ids.gen_root_id()
1203
self.root = InventoryDirectory(file_id, '', None)
1204
self._byid = {self.root.file_id: self.root}
1207
parent_path = parts[:-1]
1208
parent_id = self.path2id(parent_path)
1209
if parent_id is None:
1210
raise errors.NotVersionedError(path=parent_path)
1211
ie = make_entry(kind, parts[-1], parent_id, file_id)
450
1212
return self.add(ie)
453
1214
def __delitem__(self, file_id):
454
1215
"""Remove entry by id.
456
1217
>>> inv = Inventory()
457
>>> inv.add(InventoryEntry('123', 'foo.c', 'file', ROOT_ID))
1218
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1219
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
458
1220
>>> '123' in inv
460
1222
>>> del inv['123']
added
removed
bzrlib/inventory.py
