66
81
size in bytes of the text of the file
69
the revision_id in which the text of this file was introduced
71
83
(reading a version 4 tree created a text_id field.)
73
85
>>> i = Inventory()
76
>>> i.add(InventoryEntry('123', 'src', 'directory', ROOT_ID))
77
InventoryEntry('123', 'src', kind='directory', parent_id='TREE_ROOT')
78
>>> i.add(InventoryEntry('2323', 'hello.c', 'file', parent_id='123'))
79
InventoryEntry('2323', 'hello.c', kind='file', parent_id='123')
80
>>> for j in i.iter_entries():
88
>>> i.add(InventoryDirectory('123', 'src', ROOT_ID))
89
InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None)
90
>>> i.add(InventoryFile('2323', 'hello.c', parent_id='123'))
91
InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None)
92
>>> shouldbe = {0: '', 1: 'src', 2: 'src/hello.c'}
93
>>> for ix, j in enumerate(i.iter_entries()):
94
... print (j[0] == shouldbe[ix], j[1])
83
('src', InventoryEntry('123', 'src', kind='directory', parent_id='TREE_ROOT'))
84
('src/hello.c', InventoryEntry('2323', 'hello.c', kind='file', parent_id='123'))
85
>>> i.add(InventoryEntry('2323', 'bye.c', 'file', '123'))
96
(True, InventoryDirectory('TREE_ROOT', '', parent_id=None, revision=None))
97
(True, InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None))
98
(True, InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None))
99
>>> i.add(InventoryFile('2323', 'bye.c', '123'))
86
100
Traceback (most recent call last):
88
102
BzrError: inventory already contains entry with id {2323}
89
>>> i.add(InventoryEntry('2324', 'bye.c', 'file', '123'))
90
InventoryEntry('2324', 'bye.c', kind='file', parent_id='123')
91
>>> i.add(InventoryEntry('2325', 'wibble', 'directory', '123'))
92
InventoryEntry('2325', 'wibble', kind='directory', parent_id='123')
103
>>> i.add(InventoryFile('2324', 'bye.c', '123'))
104
InventoryFile('2324', 'bye.c', parent_id='123', sha1=None, len=None)
105
>>> i.add(InventoryDirectory('2325', 'wibble', '123'))
106
InventoryDirectory('2325', 'wibble', parent_id='123', revision=None)
93
107
>>> i.path2id('src/wibble')
97
>>> i.add(InventoryEntry('2326', 'wibble.c', 'file', '2325'))
98
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
111
>>> i.add(InventoryFile('2326', 'wibble.c', '2325'))
112
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
100
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
101
>>> for j in i.iter_entries():
103
... assert i.path2id(j[0])
114
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
115
>>> for path, entry in i.iter_entries():
117
... assert i.path2id(path)
110
125
>>> i.id2path('2326')
111
126
'src/wibble/wibble.c'
129
# Constants returned by describe_change()
131
# TODO: These should probably move to some kind of FileChangeDescription
132
# class; that's like what's inside a TreeDelta but we want to be able to
133
# generate them just for one file at a time.
135
MODIFIED_AND_RENAMED = 'modified and renamed'
114
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
115
'text_id', 'parent_id', 'children',
116
'text_version', 'entry_version', ]
119
def __init__(self, file_id, name, kind, parent_id, text_id=None):
139
def detect_changes(self, old_entry):
140
"""Return a (text_modified, meta_modified) from this to old_entry.
142
_read_tree_state must have been called on self and old_entry prior to
143
calling detect_changes.
147
def diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
148
output_to, reverse=False):
149
"""Perform a diff from this to to_entry.
151
text_diff will be used for textual difference calculation.
152
This is a template method, override _diff in child classes.
154
self._read_tree_state(tree.id2path(self.file_id), tree)
156
# cannot diff from one kind to another - you must do a removal
157
# and an addif they do not match.
158
assert self.kind == to_entry.kind
159
to_entry._read_tree_state(to_tree.id2path(to_entry.file_id),
161
self._diff(text_diff, from_label, tree, to_label, to_entry, to_tree,
164
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
165
output_to, reverse=False):
166
"""Perform a diff between two entries of the same kind."""
168
def find_previous_heads(self, previous_inventories,
169
versioned_file_store,
172
"""Return the revisions and entries that directly precede this.
174
Returned as a map from revision to inventory entry.
176
This is a map containing the file revisions in all parents
177
for which the file exists, and its revision is not a parent of
178
any other. If the file is new, the set will be empty.
180
:param versioned_file_store: A store where ancestry data on this
181
file id can be queried.
182
:param transaction: The transaction that queries to the versioned
183
file store should be completed under.
184
:param entry_vf: The entry versioned file, if its already available.
186
def get_ancestors(weave, entry):
187
return set(weave.get_ancestry(entry.revision))
188
# revision:ie mapping for each ie found in previous_inventories.
190
# revision:ie mapping with one revision for each head.
192
# revision: ancestor list for each head
194
# identify candidate head revision ids.
195
for inv in previous_inventories:
196
if self.file_id in inv:
197
ie = inv[self.file_id]
198
assert ie.file_id == self.file_id
199
if ie.revision in candidates:
200
# same revision value in two different inventories:
201
# correct possible inconsistencies:
202
# * there was a bug in revision updates with 'x' bit
205
if candidates[ie.revision].executable != ie.executable:
206
candidates[ie.revision].executable = False
207
ie.executable = False
208
except AttributeError:
210
# must now be the same.
211
assert candidates[ie.revision] == ie
213
# add this revision as a candidate.
214
candidates[ie.revision] = ie
216
# common case optimisation
217
if len(candidates) == 1:
218
# if there is only one candidate revision found
219
# then we can opening the versioned file to access ancestry:
220
# there cannot be any ancestors to eliminate when there is
221
# only one revision available.
222
heads[ie.revision] = ie
225
# eliminate ancestors amongst the available candidates:
226
# heads are those that are not an ancestor of any other candidate
227
# - this provides convergence at a per-file level.
228
for ie in candidates.values():
229
# may be an ancestor of a known head:
230
already_present = 0 != len(
231
[head for head in heads
232
if ie.revision in head_ancestors[head]])
234
# an ancestor of an analyzed candidate.
236
# not an ancestor of a known head:
237
# load the versioned file for this file id if needed
239
entry_vf = versioned_file_store.get_weave_or_empty(
240
self.file_id, transaction)
241
ancestors = get_ancestors(entry_vf, ie)
242
# may knock something else out:
243
check_heads = list(heads.keys())
244
for head in check_heads:
245
if head in ancestors:
246
# this previously discovered 'head' is not
247
# really a head - its an ancestor of the newly
250
head_ancestors[ie.revision] = ancestors
251
heads[ie.revision] = ie
254
def get_tar_item(self, root, dp, now, tree):
255
"""Get a tarfile item and a file stream for its content."""
256
item = tarfile.TarInfo(osutils.pathjoin(root, dp).encode('utf8'))
257
# TODO: would be cool to actually set it to the timestamp of the
258
# revision it was last changed
260
fileobj = self._put_in_tar(item, tree)
264
"""Return true if the object this entry represents has textual data.
266
Note that textual data includes binary content.
268
Also note that all entries get weave files created for them.
269
This attribute is primarily used when upgrading from old trees that
270
did not have the weave index for all inventory entries.
274
def __init__(self, file_id, name, parent_id, text_id=None):
120
275
"""Create an InventoryEntry
122
277
The filename must be a single component, relative to the
123
278
parent directory; it cannot be a whole path or relative name.
125
>>> e = InventoryEntry('123', 'hello.c', 'file', ROOT_ID)
280
>>> e = InventoryFile('123', 'hello.c', ROOT_ID)
130
>>> e = InventoryEntry('123', 'src/hello.c', 'file', ROOT_ID)
285
>>> e = InventoryFile('123', 'src/hello.c', ROOT_ID)
131
286
Traceback (most recent call last):
132
BzrCheckError: InventoryEntry name 'src/hello.c' is invalid
287
InvalidEntryName: Invalid entry name: src/hello.c
134
289
assert isinstance(name, basestring), name
135
290
if '/' in name or '\\' in name:
136
raise BzrCheckError('InventoryEntry name %r is invalid' % name)
138
self.text_version = None
139
self.entry_version = None
291
raise errors.InvalidEntryName(name=name)
292
self.executable = False
140
294
self.text_sha1 = None
141
295
self.text_size = None
142
296
self.file_id = file_id
145
298
self.text_id = text_id
146
299
self.parent_id = parent_id
147
if kind == 'directory':
152
raise BzrError("unhandled entry kind %r" % kind)
300
self.symlink_target = None
302
def kind_character(self):
303
"""Return a short kind indicator useful for appending to names."""
304
raise BzrError('unknown kind %r' % self.kind)
306
known_kinds = ('file', 'directory', 'symlink')
308
def _put_in_tar(self, item, tree):
309
"""populate item for stashing in a tar, and return the content stream.
311
If no content is available, return None.
313
raise BzrError("don't know how to export {%s} of kind %r" %
314
(self.file_id, self.kind))
316
def put_on_disk(self, dest, dp, tree):
317
"""Create a representation of self on disk in the prefix dest.
319
This is a template method - implement _put_on_disk in subclasses.
321
fullpath = osutils.pathjoin(dest, dp)
322
self._put_on_disk(fullpath, tree)
323
# mutter(" export {%s} kind %s to %s", self.file_id,
324
# self.kind, fullpath)
326
def _put_on_disk(self, fullpath, tree):
327
"""Put this entry onto disk at fullpath, from tree tree."""
328
raise BzrError("don't know how to export {%s} of kind %r" % (self.file_id, self.kind))
156
330
def sorted_children(self):
157
l = self.children.items()
331
return sorted(self.children.items())
334
def versionable_kind(kind):
335
return (kind in ('file', 'directory', 'symlink'))
337
def check(self, checker, rev_id, inv, tree):
338
"""Check this inventory entry is intact.
340
This is a template method, override _check for kind specific
343
:param checker: Check object providing context for the checks;
344
can be used to find out what parts of the repository have already
346
:param rev_id: Revision id from which this InventoryEntry was loaded.
347
Not necessarily the last-changed revision for this file.
348
:param inv: Inventory from which the entry was loaded.
349
:param tree: RevisionTree for this entry.
351
if self.parent_id is not None:
352
if not inv.has_id(self.parent_id):
353
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
354
% (self.parent_id, rev_id))
355
self._check(checker, rev_id, tree)
357
def _check(self, checker, rev_id, tree):
358
"""Check this inventory entry for kind specific errors."""
359
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
163
other = InventoryEntry(self.file_id, self.name, self.kind,
165
other.text_id = self.text_id
166
other.text_sha1 = self.text_sha1
167
other.text_size = self.text_size
168
other.text_version = self.text_version
169
# note that children are *not* copied; they're pulled across when
363
"""Clone this inventory entry."""
364
raise NotImplementedError
367
def describe_change(old_entry, new_entry):
368
"""Describe the change between old_entry and this.
370
This smells of being an InterInventoryEntry situation, but as its
371
the first one, we're making it a static method for now.
373
An entry with a different parent, or different name is considered
374
to be renamed. Reparenting is an internal detail.
375
Note that renaming the parent does not trigger a rename for the
378
# TODO: Perhaps return an object rather than just a string
379
if old_entry is new_entry:
380
# also the case of both being None
382
elif old_entry is None:
384
elif new_entry is None:
386
text_modified, meta_modified = new_entry.detect_changes(old_entry)
387
if text_modified or meta_modified:
391
# TODO 20060511 (mbp, rbc) factor out 'detect_rename' here.
392
if old_entry.parent_id != new_entry.parent_id:
394
elif old_entry.name != new_entry.name:
398
if renamed and not modified:
399
return InventoryEntry.RENAMED
400
if modified and not renamed:
402
if modified and renamed:
403
return InventoryEntry.MODIFIED_AND_RENAMED
174
406
def __repr__(self):
175
return ("%s(%r, %r, kind=%r, parent_id=%r)"
407
return ("%s(%r, %r, parent_id=%r, revision=%r)"
176
408
% (self.__class__.__name__,
414
def snapshot(self, revision, path, previous_entries,
415
work_tree, commit_builder):
416
"""Make a snapshot of this entry which may or may not have changed.
418
This means that all its fields are populated, that it has its
419
text stored in the text store or weave.
421
# mutter('new parents of %s are %r', path, previous_entries)
422
self._read_tree_state(path, work_tree)
423
# TODO: Where should we determine whether to reuse a
424
# previous revision id or create a new revision? 20060606
425
if len(previous_entries) == 1:
426
# cannot be unchanged unless there is only one parent file rev.
427
parent_ie = previous_entries.values()[0]
428
if self._unchanged(parent_ie):
429
# mutter("found unchanged entry")
430
self.revision = parent_ie.revision
432
return self._snapshot_into_revision(revision, previous_entries,
433
work_tree, commit_builder)
435
def _snapshot_into_revision(self, revision, previous_entries, work_tree,
437
"""Record this revision unconditionally into a store.
439
The entry's last-changed revision property (`revision`) is updated to
440
that of the new revision.
442
:param revision: id of the new revision that is being recorded.
444
:returns: String description of the commit (e.g. "merged", "modified"), etc.
446
# mutter('new revision {%s} for {%s}', revision, self.file_id)
447
self.revision = revision
448
self._snapshot_text(previous_entries, work_tree, commit_builder)
450
def _snapshot_text(self, file_parents, work_tree, commit_builder):
451
"""Record the 'text' of this entry, whatever form that takes.
453
This default implementation simply adds an empty text.
455
raise NotImplementedError(self._snapshot_text)
183
457
def __eq__(self, other):
184
458
if not isinstance(other, InventoryEntry):
185
459
return NotImplemented
187
return (self.file_id == other.file_id) \
188
and (self.name == other.name) \
189
and (self.text_sha1 == other.text_sha1) \
190
and (self.text_size == other.text_size) \
191
and (self.text_id == other.text_id) \
192
and (self.parent_id == other.parent_id) \
193
and (self.kind == other.kind) \
194
and (self.text_version == other.text_version) \
195
and (self.entry_version == other.entry_version)
461
return ((self.file_id == other.file_id)
462
and (self.name == other.name)
463
and (other.symlink_target == self.symlink_target)
464
and (self.text_sha1 == other.text_sha1)
465
and (self.text_size == other.text_size)
466
and (self.text_id == other.text_id)
467
and (self.parent_id == other.parent_id)
468
and (self.kind == other.kind)
469
and (self.revision == other.revision)
470
and (self.executable == other.executable)
198
473
def __ne__(self, other):
199
474
return not (self == other)
219
533
and (self.children == other.children)
536
class InventoryDirectory(InventoryEntry):
537
"""A directory in an inventory."""
539
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
540
'text_id', 'parent_id', 'children', 'executable',
541
'revision', 'symlink_target']
543
def _check(self, checker, rev_id, tree):
544
"""See InventoryEntry._check"""
545
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
546
raise BzrCheckError('directory {%s} has text in revision {%s}'
547
% (self.file_id, rev_id))
550
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
551
other.revision = self.revision
552
# note that children are *not* copied; they're pulled across when
556
def __init__(self, file_id, name, parent_id):
557
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
559
self.kind = 'directory'
561
def kind_character(self):
562
"""See InventoryEntry.kind_character."""
565
def _put_in_tar(self, item, tree):
566
"""See InventoryEntry._put_in_tar."""
567
item.type = tarfile.DIRTYPE
574
def _put_on_disk(self, fullpath, tree):
575
"""See InventoryEntry._put_on_disk."""
578
def _snapshot_text(self, file_parents, work_tree, commit_builder):
579
"""See InventoryEntry._snapshot_text."""
580
commit_builder.modified_directory(self.file_id, file_parents)
583
class InventoryFile(InventoryEntry):
584
"""A file in an inventory."""
586
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
587
'text_id', 'parent_id', 'children', 'executable',
588
'revision', 'symlink_target']
590
def _check(self, checker, tree_revision_id, tree):
591
"""See InventoryEntry._check"""
592
t = (self.file_id, self.revision)
593
if t in checker.checked_texts:
594
prev_sha = checker.checked_texts[t]
595
if prev_sha != self.text_sha1:
596
raise BzrCheckError('mismatched sha1 on {%s} in {%s}' %
597
(self.file_id, tree_revision_id))
599
checker.repeated_text_cnt += 1
602
if self.file_id not in checker.checked_weaves:
603
mutter('check weave {%s}', self.file_id)
604
w = tree.get_weave(self.file_id)
605
# Not passing a progress bar, because it creates a new
606
# progress, which overwrites the current progress,
607
# and doesn't look nice
609
checker.checked_weaves[self.file_id] = True
611
w = tree.get_weave(self.file_id)
613
mutter('check version {%s} of {%s}', tree_revision_id, self.file_id)
614
checker.checked_text_cnt += 1
615
# We can't check the length, because Weave doesn't store that
616
# information, and the whole point of looking at the weave's
617
# sha1sum is that we don't have to extract the text.
618
if self.text_sha1 != w.get_sha1(self.revision):
619
raise BzrCheckError('text {%s} version {%s} wrong sha1'
620
% (self.file_id, self.revision))
621
checker.checked_texts[t] = self.text_sha1
624
other = InventoryFile(self.file_id, self.name, self.parent_id)
625
other.executable = self.executable
626
other.text_id = self.text_id
627
other.text_sha1 = self.text_sha1
628
other.text_size = self.text_size
629
other.revision = self.revision
632
def detect_changes(self, old_entry):
633
"""See InventoryEntry.detect_changes."""
634
assert self.text_sha1 is not None
635
assert old_entry.text_sha1 is not None
636
text_modified = (self.text_sha1 != old_entry.text_sha1)
637
meta_modified = (self.executable != old_entry.executable)
638
return text_modified, meta_modified
640
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
641
output_to, reverse=False):
642
"""See InventoryEntry._diff."""
644
from_text = tree.get_file(self.file_id).readlines()
646
to_text = to_tree.get_file(to_entry.file_id).readlines()
650
text_diff(from_label, from_text,
651
to_label, to_text, output_to)
653
text_diff(to_label, to_text,
654
from_label, from_text, output_to)
655
except errors.BinaryFile:
657
label_pair = (to_label, from_label)
659
label_pair = (from_label, to_label)
660
print >> output_to, "Binary files %s and %s differ" % label_pair
663
"""See InventoryEntry.has_text."""
666
def __init__(self, file_id, name, parent_id):
667
super(InventoryFile, self).__init__(file_id, name, parent_id)
670
def kind_character(self):
671
"""See InventoryEntry.kind_character."""
674
def _put_in_tar(self, item, tree):
675
"""See InventoryEntry._put_in_tar."""
676
item.type = tarfile.REGTYPE
677
fileobj = tree.get_file(self.file_id)
678
item.size = self.text_size
679
if tree.is_executable(self.file_id):
685
def _put_on_disk(self, fullpath, tree):
686
"""See InventoryEntry._put_on_disk."""
687
osutils.pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
688
if tree.is_executable(self.file_id):
689
os.chmod(fullpath, 0755)
691
def _read_tree_state(self, path, work_tree):
692
"""See InventoryEntry._read_tree_state."""
693
self.text_sha1 = work_tree.get_file_sha1(self.file_id, path=path)
694
# FIXME: 20050930 probe for the text size when getting sha1
695
# in _read_tree_state
696
self.executable = work_tree.is_executable(self.file_id, path=path)
699
return ("%s(%r, %r, parent_id=%r, sha1=%r, len=%s)"
700
% (self.__class__.__name__,
707
def _forget_tree_state(self):
708
self.text_sha1 = None
710
def _snapshot_text(self, file_parents, work_tree, commit_builder):
711
"""See InventoryEntry._snapshot_text."""
712
def get_content_byte_lines():
713
return work_tree.get_file(self.file_id).readlines()
714
self.text_sha1, self.text_size = commit_builder.modified_file_text(
715
self.file_id, file_parents, get_content_byte_lines, self.text_sha1, self.text_size)
717
def _unchanged(self, previous_ie):
718
"""See InventoryEntry._unchanged."""
719
compatible = super(InventoryFile, self)._unchanged(previous_ie)
720
if self.text_sha1 != previous_ie.text_sha1:
723
# FIXME: 20050930 probe for the text size when getting sha1
724
# in _read_tree_state
725
self.text_size = previous_ie.text_size
726
if self.executable != previous_ie.executable:
731
class InventoryLink(InventoryEntry):
732
"""A file in an inventory."""
734
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
735
'text_id', 'parent_id', 'children', 'executable',
736
'revision', 'symlink_target']
738
def _check(self, checker, rev_id, tree):
739
"""See InventoryEntry._check"""
740
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
741
raise BzrCheckError('symlink {%s} has text in revision {%s}'
742
% (self.file_id, rev_id))
743
if self.symlink_target is None:
744
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
745
% (self.file_id, rev_id))
748
other = InventoryLink(self.file_id, self.name, self.parent_id)
749
other.symlink_target = self.symlink_target
750
other.revision = self.revision
753
def detect_changes(self, old_entry):
754
"""See InventoryEntry.detect_changes."""
755
# FIXME: which _modified field should we use ? RBC 20051003
756
text_modified = (self.symlink_target != old_entry.symlink_target)
758
mutter(" symlink target changed")
759
meta_modified = False
760
return text_modified, meta_modified
762
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
763
output_to, reverse=False):
764
"""See InventoryEntry._diff."""
765
from_text = self.symlink_target
766
if to_entry is not None:
767
to_text = to_entry.symlink_target
772
print >>output_to, '=== target changed %r => %r' % (from_text, to_text)
775
print >>output_to, '=== target was %r' % self.symlink_target
777
print >>output_to, '=== target is %r' % self.symlink_target
779
def __init__(self, file_id, name, parent_id):
780
super(InventoryLink, self).__init__(file_id, name, parent_id)
781
self.kind = 'symlink'
783
def kind_character(self):
784
"""See InventoryEntry.kind_character."""
787
def _put_in_tar(self, item, tree):
788
"""See InventoryEntry._put_in_tar."""
789
item.type = tarfile.SYMTYPE
793
item.linkname = self.symlink_target
796
def _put_on_disk(self, fullpath, tree):
797
"""See InventoryEntry._put_on_disk."""
799
os.symlink(self.symlink_target, fullpath)
801
raise BzrError("Failed to create symlink %r -> %r, error: %s" % (fullpath, self.symlink_target, e))
803
def _read_tree_state(self, path, work_tree):
804
"""See InventoryEntry._read_tree_state."""
805
self.symlink_target = work_tree.get_symlink_target(self.file_id)
807
def _forget_tree_state(self):
808
self.symlink_target = None
810
def _unchanged(self, previous_ie):
811
"""See InventoryEntry._unchanged."""
812
compatible = super(InventoryLink, self)._unchanged(previous_ie)
813
if self.symlink_target != previous_ie.symlink_target:
817
def _snapshot_text(self, file_parents, work_tree, commit_builder):
818
"""See InventoryEntry._snapshot_text."""
819
commit_builder.modified_link(
820
self.file_id, file_parents, self.symlink_target)
223
823
class Inventory(object):
224
824
"""Inventory of versioned files in a tree.
268
871
The inventory is created with a default root directory, with
271
# We are letting Branch(init=True) create a unique inventory
272
# root id. Rather than generating a random one here.
274
# root_id = bzrlib.branch.gen_file_id('TREE_ROOT')
275
self.root = RootEntry(root_id)
874
if root_id is not None:
875
self._set_root(InventoryDirectory(root_id, '', None))
879
self.revision_id = revision_id
881
def _set_root(self, ie):
276
883
self._byid = {self.root.file_id: self.root}
280
other = Inventory(self.root.file_id)
886
# TODO: jam 20051218 Should copy also copy the revision_id?
887
entries = self.iter_entries()
888
other = Inventory(entries.next()[1].file_id)
281
889
# copy recursively so we know directories will be added before
282
890
# their children. There are more efficient ways than this...
283
for path, entry in self.iter_entries():
284
if entry == self.root:
891
for path, entry in entries():
286
892
other.add(entry.copy())
290
895
def __iter__(self):
291
896
return iter(self._byid)
294
898
def __len__(self):
295
899
"""Returns number of entries."""
296
900
return len(self._byid)
299
902
def iter_entries(self, from_dir=None):
300
903
"""Return (path, entry) pairs, in order by name."""
304
elif isinstance(from_dir, basestring):
305
from_dir = self._byid[from_dir]
307
kids = from_dir.children.items()
309
for name, ie in kids:
311
if ie.kind == 'directory':
312
for cn, cie in self.iter_entries(from_dir=ie.file_id):
313
yield os.path.join(name, cn), cie
905
if self.root is None:
909
elif isinstance(from_dir, basestring):
910
from_dir = self._byid[from_dir]
912
# unrolling the recursive called changed the time from
913
# 440ms/663ms (inline/total) to 116ms/116ms
914
children = from_dir.children.items()
916
children = collections.deque(children)
917
stack = [(u'', children)]
919
from_dir_relpath, children = stack[-1]
922
name, ie = children.popleft()
924
# we know that from_dir_relpath never ends in a slash
925
# and 'f' doesn't begin with one, we can do a string op, rather
926
# than the checks of pathjoin(), though this means that all paths
928
path = from_dir_relpath + '/' + name
932
if ie.kind != 'directory':
935
# But do this child first
936
new_children = ie.children.items()
938
new_children = collections.deque(new_children)
939
stack.append((path, new_children))
940
# Break out of inner loop, so that we start outer loop with child
943
# if we finished all children, pop it off the stack
946
def iter_entries_by_dir(self, from_dir=None):
947
"""Iterate over the entries in a directory first order.
949
This returns all entries for a directory before returning
950
the entries for children of a directory. This is not
951
lexicographically sorted order, and is a hybrid between
952
depth-first and breadth-first.
954
:return: This yields (path, entry) pairs
956
# TODO? Perhaps this should return the from_dir so that the root is
957
# yielded? or maybe an option?
959
if self.root is None:
963
elif isinstance(from_dir, basestring):
964
from_dir = self._byid[from_dir]
966
stack = [(u'', from_dir)]
968
cur_relpath, cur_dir = stack.pop()
971
for child_name, child_ie in sorted(cur_dir.children.iteritems()):
973
child_relpath = cur_relpath + child_name
975
yield child_relpath, child_ie
977
if child_ie.kind == 'directory':
978
child_dirs.append((child_relpath+'/', child_ie))
979
stack.extend(reversed(child_dirs))
316
981
def entries(self):
317
982
"""Return list of (path, ie) for all entries except the root.
added
removed
bzrlib/inventory.py
