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:
573
raise BzrCheckError('mismatched sha1 on {%s} in {%s}' %
574
(self.file_id, tree_revision_id))
576
checker.repeated_text_cnt += 1
579
if self.file_id not in checker.checked_weaves:
580
mutter('check weave {%s}', self.file_id)
581
w = tree._get_weave(self.file_id)
582
# Not passing a progress bar, because it creates a new
583
# progress, which overwrites the current progress,
584
# and doesn't look nice
586
checker.checked_weaves[self.file_id] = True
588
w = tree._get_weave(self.file_id)
590
mutter('check version {%s} of {%s}', tree_revision_id, self.file_id)
591
checker.checked_text_cnt += 1
592
# We can't check the length, because Weave doesn't store that
593
# information, and the whole point of looking at the weave's
594
# sha1sum is that we don't have to extract the text.
595
if self.text_sha1 != w.get_sha1(self.revision):
596
raise BzrCheckError('text {%s} version {%s} wrong sha1'
597
% (self.file_id, self.revision))
598
checker.checked_texts[t] = self.text_sha1
601
other = InventoryFile(self.file_id, self.name, self.parent_id)
602
other.executable = self.executable
603
other.text_id = self.text_id
604
other.text_sha1 = self.text_sha1
605
other.text_size = self.text_size
606
other.revision = self.revision
609
def detect_changes(self, old_entry):
610
"""See InventoryEntry.detect_changes."""
611
assert self.text_sha1 is not None
612
assert old_entry.text_sha1 is not None
613
text_modified = (self.text_sha1 != old_entry.text_sha1)
614
meta_modified = (self.executable != old_entry.executable)
615
return text_modified, meta_modified
617
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
618
output_to, reverse=False):
619
"""See InventoryEntry._diff."""
621
from_text = tree.get_file(self.file_id).readlines()
623
to_text = to_tree.get_file(to_entry.file_id).readlines()
627
text_diff(from_label, from_text,
628
to_label, to_text, output_to)
630
text_diff(to_label, to_text,
631
from_label, from_text, output_to)
632
except errors.BinaryFile:
634
label_pair = (to_label, from_label)
636
label_pair = (from_label, to_label)
637
print >> output_to, \
638
("Binary files %s and %s differ" % label_pair).encode('utf8')
641
"""See InventoryEntry.has_text."""
644
def __init__(self, file_id, name, parent_id):
645
super(InventoryFile, self).__init__(file_id, name, parent_id)
648
def kind_character(self):
649
"""See InventoryEntry.kind_character."""
652
def _put_in_tar(self, item, tree):
653
"""See InventoryEntry._put_in_tar."""
654
item.type = tarfile.REGTYPE
655
fileobj = tree.get_file(self.file_id)
656
item.size = self.text_size
657
if tree.is_executable(self.file_id):
663
def _put_on_disk(self, fullpath, tree):
664
"""See InventoryEntry._put_on_disk."""
665
osutils.pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
666
if tree.is_executable(self.file_id):
667
os.chmod(fullpath, 0755)
669
def _read_tree_state(self, path, work_tree):
670
"""See InventoryEntry._read_tree_state."""
671
self.text_sha1 = work_tree.get_file_sha1(self.file_id, path=path)
672
# FIXME: 20050930 probe for the text size when getting sha1
673
# in _read_tree_state
674
self.executable = work_tree.is_executable(self.file_id, path=path)
677
return ("%s(%r, %r, parent_id=%r, sha1=%r, len=%s)"
678
% (self.__class__.__name__,
685
def _forget_tree_state(self):
686
self.text_sha1 = None
688
def _unchanged(self, previous_ie):
689
"""See InventoryEntry._unchanged."""
690
compatible = super(InventoryFile, self)._unchanged(previous_ie)
691
if self.text_sha1 != previous_ie.text_sha1:
694
# FIXME: 20050930 probe for the text size when getting sha1
695
# in _read_tree_state
696
self.text_size = previous_ie.text_size
697
if self.executable != previous_ie.executable:
702
class InventoryLink(InventoryEntry):
703
"""A file in an inventory."""
705
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
706
'text_id', 'parent_id', 'children', 'executable',
707
'revision', 'symlink_target', 'reference_revision']
709
def _check(self, checker, rev_id, tree):
710
"""See InventoryEntry._check"""
711
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
712
raise BzrCheckError('symlink {%s} has text in revision {%s}'
713
% (self.file_id, rev_id))
714
if self.symlink_target is None:
715
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
716
% (self.file_id, rev_id))
719
other = InventoryLink(self.file_id, self.name, self.parent_id)
720
other.symlink_target = self.symlink_target
721
other.revision = self.revision
724
def detect_changes(self, old_entry):
725
"""See InventoryEntry.detect_changes."""
726
# FIXME: which _modified field should we use ? RBC 20051003
727
text_modified = (self.symlink_target != old_entry.symlink_target)
729
mutter(" symlink target changed")
730
meta_modified = False
731
return text_modified, meta_modified
733
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
734
output_to, reverse=False):
735
"""See InventoryEntry._diff."""
736
from_text = self.symlink_target
737
if to_entry is not None:
738
to_text = to_entry.symlink_target
743
print >>output_to, '=== target changed %r => %r' % (from_text, to_text)
746
print >>output_to, '=== target was %r' % self.symlink_target
748
print >>output_to, '=== target is %r' % self.symlink_target
750
def __init__(self, file_id, name, parent_id):
751
super(InventoryLink, self).__init__(file_id, name, parent_id)
752
self.kind = 'symlink'
754
def kind_character(self):
755
"""See InventoryEntry.kind_character."""
758
def _put_in_tar(self, item, tree):
759
"""See InventoryEntry._put_in_tar."""
760
item.type = tarfile.SYMTYPE
764
item.linkname = self.symlink_target
767
def _put_on_disk(self, fullpath, tree):
768
"""See InventoryEntry._put_on_disk."""
770
os.symlink(self.symlink_target, fullpath)
772
raise BzrError("Failed to create symlink %r -> %r, error: %s" % (fullpath, self.symlink_target, e))
774
def _read_tree_state(self, path, work_tree):
775
"""See InventoryEntry._read_tree_state."""
776
self.symlink_target = work_tree.get_symlink_target(self.file_id)
778
def _forget_tree_state(self):
779
self.symlink_target = None
781
def _unchanged(self, previous_ie):
782
"""See InventoryEntry._unchanged."""
783
compatible = super(InventoryLink, self)._unchanged(previous_ie)
784
if self.symlink_target != previous_ie.symlink_target:
789
class TreeReference(InventoryEntry):
791
kind = 'tree-reference'
793
def __init__(self, file_id, name, parent_id, revision=None,
794
reference_revision=None):
795
InventoryEntry.__init__(self, file_id, name, parent_id)
796
self.revision = revision
797
self.reference_revision = reference_revision
800
return TreeReference(self.file_id, self.name, self.parent_id,
801
self.revision, self.reference_revision)
803
def _read_tree_state(self, path, work_tree):
804
"""Populate fields in the inventory entry from the given tree.
806
self.reference_revision = work_tree.get_reference_revision(
809
def _forget_tree_state(self):
810
self.reference_revision = None
812
def _unchanged(self, previous_ie):
813
"""See InventoryEntry._unchanged."""
814
compatible = super(TreeReference, self)._unchanged(previous_ie)
815
if self.reference_revision != previous_ie.reference_revision:
252
820
class Inventory(object):
253
821
"""Inventory of versioned files in a tree.
295
868
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)
871
if root_id is not None:
872
assert root_id.__class__ == str
873
self._set_root(InventoryDirectory(root_id, u'', None))
877
self.revision_id = revision_id
879
def _set_root(self, ie):
303
881
self._byid = {self.root.file_id: self.root}
884
# TODO: jam 20051218 Should copy also copy the revision_id?
885
entries = self.iter_entries()
886
other = Inventory(entries.next()[1].file_id)
887
# copy recursively so we know directories will be added before
888
# their children. There are more efficient ways than this...
889
for path, entry in entries():
890
other.add(entry.copy())
306
893
def __iter__(self):
307
894
return iter(self._byid)
310
896
def __len__(self):
311
897
"""Returns number of entries."""
312
898
return len(self._byid)
315
900
def iter_entries(self, from_dir=None):
316
901
"""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
903
if self.root is None:
907
elif isinstance(from_dir, basestring):
908
from_dir = self._byid[from_dir]
910
# unrolling the recursive called changed the time from
911
# 440ms/663ms (inline/total) to 116ms/116ms
912
children = from_dir.children.items()
914
children = collections.deque(children)
915
stack = [(u'', children)]
917
from_dir_relpath, children = stack[-1]
920
name, ie = children.popleft()
922
# we know that from_dir_relpath never ends in a slash
923
# and 'f' doesn't begin with one, we can do a string op, rather
924
# than the checks of pathjoin(), though this means that all paths
926
path = from_dir_relpath + '/' + name
930
if ie.kind != 'directory':
933
# But do this child first
934
new_children = ie.children.items()
936
new_children = collections.deque(new_children)
937
stack.append((path, new_children))
938
# Break out of inner loop, so that we start outer loop with child
941
# if we finished all children, pop it off the stack
944
def iter_entries_by_dir(self, from_dir=None, specific_file_ids=None,
945
yield_parents=False):
946
"""Iterate over the entries in a directory first order.
948
This returns all entries for a directory before returning
949
the entries for children of a directory. This is not
950
lexicographically sorted order, and is a hybrid between
951
depth-first and breadth-first.
953
:param yield_parents: If True, yield the parents from the root leading
954
down to specific_file_ids that have been requested. This has no
955
impact if specific_file_ids is None.
956
:return: This yields (path, entry) pairs
958
if specific_file_ids:
959
safe = osutils.safe_file_id
960
specific_file_ids = set(safe(fid) for fid in specific_file_ids)
961
# TODO? Perhaps this should return the from_dir so that the root is
962
# yielded? or maybe an option?
964
if self.root is None:
966
# Optimize a common case
967
if (not yield_parents and specific_file_ids is not None and
968
len(specific_file_ids) == 1):
969
file_id = list(specific_file_ids)[0]
971
yield self.id2path(file_id), self[file_id]
974
if (specific_file_ids is None or yield_parents or
975
self.root.file_id in specific_file_ids):
977
elif isinstance(from_dir, basestring):
978
from_dir = self._byid[from_dir]
980
if specific_file_ids is not None:
981
# TODO: jam 20070302 This could really be done as a loop rather
982
# than a bunch of recursive calls.
985
def add_ancestors(file_id):
986
if file_id not in byid:
988
parent_id = byid[file_id].parent_id
989
if parent_id is None:
991
if parent_id not in parents:
992
parents.add(parent_id)
993
add_ancestors(parent_id)
994
for file_id in specific_file_ids:
995
add_ancestors(file_id)
999
stack = [(u'', from_dir)]
1001
cur_relpath, cur_dir = stack.pop()
1004
for child_name, child_ie in sorted(cur_dir.children.iteritems()):
1006
child_relpath = cur_relpath + child_name
1008
if (specific_file_ids is None or
1009
child_ie.file_id in specific_file_ids or
1010
(yield_parents and child_ie.file_id in parents)):
1011
yield child_relpath, child_ie
1013
if child_ie.kind == 'directory':
1014
if parents is None or child_ie.file_id in parents:
1015
child_dirs.append((child_relpath+'/', child_ie))
1016
stack.extend(reversed(child_dirs))
1018
def make_entry(self, kind, name, parent_id, file_id=None):
1019
"""Simple thunk to bzrlib.inventory.make_entry."""
1020
return make_entry(kind, name, parent_id, file_id)
332
1022
def entries(self):
333
1023
"""Return list of (path, ie) for all entries except the root.
361
1050
for name, child_ie in kids:
362
child_path = os.path.join(parent_path, name)
1051
child_path = osutils.pathjoin(parent_path, name)
363
1052
descend(child_ie, child_path)
364
descend(self.root, '')
1053
descend(self.root, u'')
369
1056
def __contains__(self, file_id):
370
1057
"""True if this entry contains a file with given id.
372
1059
>>> inv = Inventory()
373
>>> inv.add(InventoryEntry('123', 'foo.c', 'file', ROOT_ID))
1060
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1061
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
374
1062
>>> '123' in inv
376
1064
>>> '456' in inv
379
return file_id in self._byid
1067
file_id = osutils.safe_file_id(file_id)
1068
return (file_id in self._byid)
382
1070
def __getitem__(self, file_id):
383
1071
"""Return the entry for given file_id.
385
1073
>>> inv = Inventory()
386
>>> inv.add(InventoryEntry('123123', 'hello.c', 'file', ROOT_ID))
1074
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
1075
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
387
1076
>>> inv['123123'].name
1079
file_id = osutils.safe_file_id(file_id)
391
1081
return self._byid[file_id]
392
1082
except KeyError:
394
raise BzrError("can't look up file_id None")
396
raise BzrError("file_id {%s} not in inventory" % file_id)
1083
# really we're passing an inventory, not a tree...
1084
raise errors.NoSuchId(self, file_id)
399
1086
def get_file_kind(self, file_id):
1087
file_id = osutils.safe_file_id(file_id)
400
1088
return self._byid[file_id].kind
402
1090
def get_child(self, parent_id, filename):
1091
parent_id = osutils.safe_file_id(parent_id)
403
1092
return self[parent_id].children.get(filename)
1094
def _add_child(self, entry):
1095
"""Add an entry to the inventory, without adding it to its parent"""
1096
if entry.file_id in self._byid:
1097
raise BzrError("inventory already contains entry with id {%s}" %
1099
self._byid[entry.file_id] = entry
1100
for child in getattr(entry, 'children', {}).itervalues():
1101
self._add_child(child)
406
1104
def add(self, entry):
407
1105
"""Add entry to inventory.
409
1107
To add a file to a branch ready to be committed, use Branch.add,
1110
Returns the new entry object.
411
1112
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):
1113
raise errors.DuplicateFileId(entry.file_id,
1114
self._byid[entry.file_id])
1116
if entry.parent_id is None:
1117
assert self.root is None and len(self._byid) == 0
1121
parent = self._byid[entry.parent_id]
1123
raise BzrError("parent_id {%s} not in inventory" %
1126
if entry.name in parent.children:
1127
raise BzrError("%s is already versioned" %
1128
osutils.pathjoin(self.id2path(parent.file_id),
1129
entry.name).encode('utf-8'))
1130
parent.children[entry.name] = entry
1131
return self._add_child(entry)
1133
def add_path(self, relpath, kind, file_id=None, parent_id=None):
431
1134
"""Add entry from a path.
433
The immediate parent must already be versioned"""
434
from bzrlib.branch import gen_file_id
1136
The immediate parent must already be versioned.
1138
Returns the new entry object."""
436
parts = bzrlib.osutils.splitpath(relpath)
1140
parts = osutils.splitpath(relpath)
437
1142
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)
1144
file_id = generate_ids.gen_root_id()
1146
file_id = osutils.safe_file_id(file_id)
1147
self.root = InventoryDirectory(file_id, '', None)
1148
self._byid = {self.root.file_id: self.root}
1151
parent_path = parts[:-1]
1152
parent_id = self.path2id(parent_path)
1153
if parent_id is None:
1154
raise errors.NotVersionedError(path=parent_path)
1155
ie = make_entry(kind, parts[-1], parent_id, file_id)
450
1156
return self.add(ie)
453
1158
def __delitem__(self, file_id):
454
1159
"""Remove entry by id.
456
1161
>>> inv = Inventory()
457
>>> inv.add(InventoryEntry('123', 'foo.c', 'file', ROOT_ID))
1162
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1163
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
458
1164
>>> '123' in inv
460
1166
>>> del inv['123']
461
1167
>>> '123' in inv
1170
file_id = osutils.safe_file_id(file_id)
464
1171
ie = self[file_id]
466
assert self[ie.parent_id].children[ie.name] == ie
1173
assert ie.parent_id is None or \
1174
self[ie.parent_id].children[ie.name] == ie
468
# TODO: Test deleting all children; maybe hoist to a separate
470
if ie.kind == 'directory':
471
for cie in ie.children.values():
472
del self[cie.file_id]
475
1176
del self._byid[file_id]
476
del self[ie.parent_id].children[ie.name]
479
def to_element(self):
480
"""Convert to XML Element"""
481
from bzrlib.xml import Element
483
e = Element('inventory')
485
if self.root.file_id not in (None, ROOT_ID):
486
e.set('file_id', self.root.file_id)
487
for path, ie in self.iter_entries():
488
e.append(ie.to_element())
492
def from_element(cls, elt):
493
"""Construct from XML Element
495
>>> inv = Inventory()
496
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c', 'file', ROOT_ID))
497
>>> elt = inv.to_element()
498
>>> inv2 = Inventory.from_element(elt)
502
# XXXX: doctest doesn't run this properly under python2.3
503
assert elt.tag == 'inventory'
504
root_id = elt.get('file_id') or ROOT_ID
507
ie = InventoryEntry.from_element(e)
508
if ie.parent_id == ROOT_ID:
509
ie.parent_id = root_id
513
from_element = classmethod(from_element)
1177
if ie.parent_id is not None:
1178
del self[ie.parent_id].children[ie.name]
516
1180
def __eq__(self, other):
517
1181
"""Compare two sets by comparing their contents.
added
removed
bzrlib/inventory.py
