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():
85
121
src/wibble/wibble.c
86
122
>>> i.id2path('2326')
87
123
'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):
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
def get_tar_item(self, root, dp, now, tree):
180
"""Get a tarfile item and a file stream for its content."""
181
item = tarfile.TarInfo(osutils.pathjoin(root, dp).encode('utf8'))
182
# TODO: would be cool to actually set it to the timestamp of the
183
# revision it was last changed
185
fileobj = self._put_in_tar(item, tree)
189
"""Return true if the object this entry represents has textual data.
191
Note that textual data includes binary content.
193
Also note that all entries get weave files created for them.
194
This attribute is primarily used when upgrading from old trees that
195
did not have the weave index for all inventory entries.
199
def __init__(self, file_id, name, parent_id, text_id=None):
101
200
"""Create an InventoryEntry
103
202
The filename must be a single component, relative to the
104
203
parent directory; it cannot be a whole path or relative name.
106
>>> e = InventoryEntry('123', 'hello.c', 'file', ROOT_ID)
205
>>> e = InventoryFile('123', 'hello.c', ROOT_ID)
111
>>> e = InventoryEntry('123', 'src/hello.c', 'file', ROOT_ID)
210
>>> e = InventoryFile('123', 'src/hello.c', ROOT_ID)
112
211
Traceback (most recent call last):
113
BzrCheckError: InventoryEntry name 'src/hello.c' is invalid
212
InvalidEntryName: Invalid entry name: src/hello.c
115
214
if '/' in name or '\\' in name:
116
raise BzrCheckError('InventoryEntry name %r is invalid' % name)
215
raise errors.InvalidEntryName(name=name)
216
self.executable = False
118
218
self.text_sha1 = None
119
219
self.text_size = None
121
220
self.file_id = file_id
124
222
self.text_id = text_id
125
223
self.parent_id = parent_id
126
if kind == 'directory':
131
raise BzrError("unhandled entry kind %r" % kind)
224
self.symlink_target = None
225
self.reference_revision = None
227
def kind_character(self):
228
"""Return a short kind indicator useful for appending to names."""
229
raise BzrError('unknown kind %r' % self.kind)
231
known_kinds = ('file', 'directory', 'symlink')
233
def _put_in_tar(self, item, tree):
234
"""populate item for stashing in a tar, and return the content stream.
236
If no content is available, return None.
238
raise BzrError("don't know how to export {%s} of kind %r" %
239
(self.file_id, self.kind))
241
def put_on_disk(self, dest, dp, tree):
242
"""Create a representation of self on disk in the prefix dest.
244
This is a template method - implement _put_on_disk in subclasses.
246
fullpath = osutils.pathjoin(dest, dp)
247
self._put_on_disk(fullpath, tree)
248
# mutter(" export {%s} kind %s to %s", self.file_id,
249
# self.kind, fullpath)
251
def _put_on_disk(self, fullpath, tree):
252
"""Put this entry onto disk at fullpath, from tree tree."""
253
raise BzrError("don't know how to export {%s} of kind %r" % (self.file_id, self.kind))
135
255
def sorted_children(self):
136
l = self.children.items()
256
return sorted(self.children.items())
259
def versionable_kind(kind):
260
return (kind in ('file', 'directory', 'symlink', 'tree-reference'))
262
def check(self, checker, rev_id, inv, tree):
263
"""Check this inventory entry is intact.
265
This is a template method, override _check for kind specific
268
:param checker: Check object providing context for the checks;
269
can be used to find out what parts of the repository have already
271
:param rev_id: Revision id from which this InventoryEntry was loaded.
272
Not necessarily the last-changed revision for this file.
273
:param inv: Inventory from which the entry was loaded.
274
:param tree: RevisionTree for this entry.
276
if self.parent_id is not None:
277
if not inv.has_id(self.parent_id):
278
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
279
% (self.parent_id, rev_id))
280
self._check(checker, rev_id, tree)
282
def _check(self, checker, rev_id, tree):
283
"""Check this inventory entry for kind specific errors."""
284
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
288
"""Clone this inventory entry."""
289
raise NotImplementedError
292
def describe_change(old_entry, new_entry):
293
"""Describe the change between old_entry and this.
295
This smells of being an InterInventoryEntry situation, but as its
296
the first one, we're making it a static method for now.
298
An entry with a different parent, or different name is considered
299
to be renamed. Reparenting is an internal detail.
300
Note that renaming the parent does not trigger a rename for the
303
# TODO: Perhaps return an object rather than just a string
304
if old_entry is new_entry:
305
# also the case of both being None
307
elif old_entry is None:
309
elif new_entry is None:
311
if old_entry.kind != new_entry.kind:
313
text_modified, meta_modified = new_entry.detect_changes(old_entry)
314
if text_modified or meta_modified:
318
# TODO 20060511 (mbp, rbc) factor out 'detect_rename' here.
319
if old_entry.parent_id != new_entry.parent_id:
321
elif old_entry.name != new_entry.name:
325
if renamed and not modified:
326
return InventoryEntry.RENAMED
327
if modified and not renamed:
329
if modified and renamed:
330
return InventoryEntry.MODIFIED_AND_RENAMED
151
333
def __repr__(self):
152
return ("%s(%r, %r, kind=%r, parent_id=%r)"
334
return ("%s(%r, %r, parent_id=%r, revision=%r)"
153
335
% (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
341
def __eq__(self, other):
215
342
if not isinstance(other, InventoryEntry):
216
343
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)
345
return ((self.file_id == other.file_id)
346
and (self.name == other.name)
347
and (other.symlink_target == self.symlink_target)
348
and (self.text_sha1 == other.text_sha1)
349
and (self.text_size == other.text_size)
350
and (self.text_id == other.text_id)
351
and (self.parent_id == other.parent_id)
352
and (self.kind == other.kind)
353
and (self.revision == other.revision)
354
and (self.executable == other.executable)
355
and (self.reference_revision == other.reference_revision)
227
358
def __ne__(self, other):
228
359
return not (self == other)
248
420
and (self.children == other.children)
423
class InventoryDirectory(InventoryEntry):
424
"""A directory in an inventory."""
426
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
427
'text_id', 'parent_id', 'children', 'executable',
428
'revision', 'symlink_target', 'reference_revision']
430
def _check(self, checker, rev_id, tree):
431
"""See InventoryEntry._check"""
432
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
433
raise BzrCheckError('directory {%s} has text in revision {%s}'
434
% (self.file_id, rev_id))
437
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
438
other.revision = self.revision
439
# note that children are *not* copied; they're pulled across when
443
def __init__(self, file_id, name, parent_id):
444
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
446
self.kind = 'directory'
448
def kind_character(self):
449
"""See InventoryEntry.kind_character."""
452
def _put_in_tar(self, item, tree):
453
"""See InventoryEntry._put_in_tar."""
454
item.type = tarfile.DIRTYPE
461
def _put_on_disk(self, fullpath, tree):
462
"""See InventoryEntry._put_on_disk."""
466
class InventoryFile(InventoryEntry):
467
"""A file in an inventory."""
469
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
470
'text_id', 'parent_id', 'children', 'executable',
471
'revision', 'symlink_target', 'reference_revision']
473
def _check(self, checker, tree_revision_id, tree):
474
"""See InventoryEntry._check"""
475
t = (self.file_id, self.revision)
476
if t in checker.checked_texts:
477
prev_sha = checker.checked_texts[t]
478
if prev_sha != self.text_sha1:
480
'mismatched sha1 on {%s} in {%s} (%s != %s) %r' %
481
(self.file_id, tree_revision_id, prev_sha, self.text_sha1,
484
checker.repeated_text_cnt += 1
487
mutter('check version {%s} of {%s}', tree_revision_id, self.file_id)
488
checker.checked_text_cnt += 1
489
# We can't check the length, because Weave doesn't store that
490
# information, and the whole point of looking at the weave's
491
# sha1sum is that we don't have to extract the text.
492
if (self.text_sha1 !=
493
tree._repository.texts.get_sha1s([(self.file_id, self.revision)])[0]):
494
raise BzrCheckError('text {%s} version {%s} wrong sha1'
495
% (self.file_id, self.revision))
496
checker.checked_texts[t] = 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:
252
712
class Inventory(object):
253
713
"""Inventory of versioned files in a tree.
295
760
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)
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):
303
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())
306
849
def __iter__(self):
307
850
return iter(self._byid)
310
852
def __len__(self):
311
853
"""Returns number of entries."""
312
854
return len(self._byid)
315
856
def iter_entries(self, from_dir=None):
316
857
"""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
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)
332
977
def entries(self):
333
978
"""Return list of (path, ie) for all entries except the root.
402
1042
def get_child(self, parent_id, filename):
403
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)
406
1055
def add(self, entry):
407
1056
"""Add entry to inventory.
409
1058
To add a file to a branch ready to be committed, use Branch.add,
1061
Returns the new entry object.
411
1063
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):
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):
431
1084
"""Add entry from a path.
433
The immediate parent must already be versioned"""
434
from bzrlib.branch import gen_file_id
1086
The immediate parent must already be versioned.
1088
Returns the new entry object."""
436
parts = bzrlib.osutils.splitpath(relpath)
1090
parts = osutils.splitpath(relpath)
437
1092
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)
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)
450
1104
return self.add(ie)
453
1106
def __delitem__(self, file_id):
454
1107
"""Remove entry by id.
456
1109
>>> inv = Inventory()
457
>>> inv.add(InventoryEntry('123', 'foo.c', 'file', ROOT_ID))
1110
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1111
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
458
1112
>>> '123' in inv
460
1114
>>> del inv['123']
added
removed
bzrlib/inventory.py
