78
89
>>> i.add(InventoryDirectory('123', 'src', ROOT_ID))
79
InventoryDirectory('123', 'src', parent_id='TREE_ROOT')
90
InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None)
80
91
>>> i.add(InventoryFile('2323', 'hello.c', parent_id='123'))
81
InventoryFile('2323', 'hello.c', parent_id='123')
82
>>> shouldbe = {0: 'src', 1: os.path.join('src','hello.c')}
92
InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None)
93
>>> shouldbe = {0: '', 1: 'src', 2: 'src/hello.c'}
83
94
>>> for ix, j in enumerate(i.iter_entries()):
84
95
... print (j[0] == shouldbe[ix], j[1])
86
(True, InventoryDirectory('123', 'src', parent_id='TREE_ROOT'))
87
(True, InventoryFile('2323', 'hello.c', parent_id='123'))
88
>>> i.add(InventoryFile('2323', 'bye.c', '123'))
89
Traceback (most recent call last):
91
BzrError: inventory already contains entry with id {2323}
97
(True, InventoryDirectory('TREE_ROOT', u'', parent_id=None, revision=None))
98
(True, InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None))
99
(True, InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None))
92
100
>>> i.add(InventoryFile('2324', 'bye.c', '123'))
93
InventoryFile('2324', 'bye.c', parent_id='123')
101
InventoryFile('2324', 'bye.c', parent_id='123', sha1=None, len=None)
94
102
>>> i.add(InventoryDirectory('2325', 'wibble', '123'))
95
InventoryDirectory('2325', 'wibble', parent_id='123')
103
InventoryDirectory('2325', 'wibble', parent_id='123', revision=None)
96
104
>>> i.path2id('src/wibble')
100
108
>>> i.add(InventoryFile('2326', 'wibble.c', '2325'))
101
InventoryFile('2326', 'wibble.c', parent_id='2325')
109
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
103
InventoryFile('2326', 'wibble.c', parent_id='2325')
111
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
104
112
>>> for path, entry in i.iter_entries():
105
... print path.replace('\\\\', '/') # for win32 os.sep
106
114
... assert i.path2id(path)
112
121
src/wibble/wibble.c
113
>>> i.id2path('2326').replace('\\\\', '/')
122
>>> i.id2path('2326')
114
123
'src/wibble/wibble.c'
126
# Constants returned by describe_change()
128
# TODO: These should probably move to some kind of FileChangeDescription
129
# class; that's like what's inside a TreeDelta but we want to be able to
130
# generate them just for one file at a time.
132
MODIFIED_AND_RENAMED = 'modified and renamed'
117
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
118
'text_id', 'parent_id', 'children', 'executable',
121
def _add_text_to_weave(self, new_lines, parents, weave_store, transaction):
122
weave_store.add_text(self.file_id, self.revision, new_lines, parents,
125
136
def detect_changes(self, old_entry):
126
137
"""Return a (text_modified, meta_modified) from this to old_entry.
151
162
output_to, reverse=False):
152
163
"""Perform a diff between two entries of the same kind."""
154
def find_previous_heads(self, previous_inventories, entry_weave):
155
"""Return the revisions and entries that directly preceed this.
165
def find_previous_heads(self, previous_inventories,
166
versioned_file_store,
169
"""Return the revisions and entries that directly precede this.
157
171
Returned as a map from revision to inventory entry.
159
173
This is a map containing the file revisions in all parents
160
174
for which the file exists, and its revision is not a parent of
161
175
any other. If the file is new, the set will be empty.
177
:param versioned_file_store: A store where ancestry data on this
178
file id can be queried.
179
:param transaction: The transaction that queries to the versioned
180
file store should be completed under.
181
:param entry_vf: The entry versioned file, if its already available.
163
183
def get_ancestors(weave, entry):
164
return set(map(weave.idx_to_name,
165
weave.inclusions([weave.lookup(entry.revision)])))
184
return set(weave.get_ancestry(entry.revision))
185
# revision:ie mapping for each ie found in previous_inventories.
187
# revision:ie mapping with one revision for each head.
189
# revision: ancestor list for each head
167
190
head_ancestors = {}
191
# identify candidate head revision ids.
168
192
for inv in previous_inventories:
169
193
if self.file_id in inv:
170
194
ie = inv[self.file_id]
171
195
assert ie.file_id == self.file_id
172
if ie.revision in heads:
173
# fixup logic, there was a bug in revision updates.
174
# with x bit support.
196
if ie.kind != self.kind:
197
# Can't be a candidate if the kind has changed.
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
176
if heads[ie.revision].executable != ie.executable:
177
heads[ie.revision].executable = False
205
if candidates[ie.revision].executable != ie.executable:
206
candidates[ie.revision].executable = False
178
207
ie.executable = False
179
208
except AttributeError:
181
assert heads[ie.revision] == ie
210
# must now be the same.
211
assert candidates[ie.revision] == ie
183
# may want to add it.
184
# may already be covered:
185
already_present = 0 != len(
186
[head for head in heads
187
if ie.revision in head_ancestors[head]])
189
# an ancestor of a known head.
192
ancestors = get_ancestors(entry_weave, ie)
193
# may knock something else out:
194
check_heads = list(heads.keys())
195
for head in check_heads:
196
if head in ancestors:
197
# this head is not really a head
199
head_ancestors[ie.revision] = ancestors
200
heads[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
203
254
def get_tar_item(self, root, dp, now, tree):
204
255
"""Get a tarfile item and a file stream for its content."""
205
item = tarfile.TarInfo(os.path.join(root, dp))
256
item = tarfile.TarInfo(osutils.pathjoin(root, dp).encode('utf8'))
206
257
# TODO: would be cool to actually set it to the timestamp of the
207
258
# revision it was last changed
238
289
assert isinstance(name, basestring), name
239
290
if '/' in name or '\\' in name:
240
raise InvalidEntryName(name=name)
291
raise errors.InvalidEntryName(name=name)
241
292
self.executable = False
242
293
self.revision = None
243
294
self.text_sha1 = None
244
295
self.text_size = None
245
296
self.file_id = file_id
297
assert isinstance(file_id, (str, None.__class__)), \
298
'bad type %r for %r' % (type(file_id), file_id)
247
300
self.text_id = text_id
248
301
self.parent_id = parent_id
249
302
self.symlink_target = None
303
self.reference_revision = None
251
305
def kind_character(self):
252
306
"""Return a short kind indicator useful for appending to names."""
253
307
raise BzrError('unknown kind %r' % self.kind)
255
known_kinds = ('file', 'directory', 'symlink', 'root_directory')
309
known_kinds = ('file', 'directory', 'symlink')
257
311
def _put_in_tar(self, item, tree):
258
312
"""populate item for stashing in a tar, and return the content stream.
268
322
This is a template method - implement _put_on_disk in subclasses.
270
fullpath = appendpath(dest, dp)
324
fullpath = osutils.pathjoin(dest, dp)
271
325
self._put_on_disk(fullpath, tree)
272
mutter(" export {%s} kind %s to %s", self.file_id,
326
# mutter(" export {%s} kind %s to %s", self.file_id,
327
# self.kind, fullpath)
275
329
def _put_on_disk(self, fullpath, tree):
276
330
"""Put this entry onto disk at fullpath, from tree tree."""
277
331
raise BzrError("don't know how to export {%s} of kind %r" % (self.file_id, self.kind))
279
333
def sorted_children(self):
280
l = self.children.items()
334
return sorted(self.children.items())
285
337
def versionable_kind(kind):
286
return kind in ('file', 'directory', 'symlink')
338
return (kind in ('file', 'directory', 'symlink', 'tree-reference'))
288
340
def check(self, checker, rev_id, inv, tree):
289
341
"""Check this inventory entry is intact.
291
343
This is a template method, override _check for kind specific
346
:param checker: Check object providing context for the checks;
347
can be used to find out what parts of the repository have already
349
:param rev_id: Revision id from which this InventoryEntry was loaded.
350
Not necessarily the last-changed revision for this file.
351
:param inv: Inventory from which the entry was loaded.
352
:param tree: RevisionTree for this entry.
294
if self.parent_id != None:
354
if self.parent_id is not None:
295
355
if not inv.has_id(self.parent_id):
296
356
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
297
357
% (self.parent_id, rev_id))
302
362
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
303
363
(self.kind, rev_id))
307
366
"""Clone this inventory entry."""
308
367
raise NotImplementedError
310
def _get_snapshot_change(self, previous_entries):
311
if len(previous_entries) > 1:
313
elif len(previous_entries) == 0:
370
def describe_change(old_entry, new_entry):
371
"""Describe the change between old_entry and this.
373
This smells of being an InterInventoryEntry situation, but as its
374
the first one, we're making it a static method for now.
376
An entry with a different parent, or different name is considered
377
to be renamed. Reparenting is an internal detail.
378
Note that renaming the parent does not trigger a rename for the
381
# TODO: Perhaps return an object rather than just a string
382
if old_entry is new_entry:
383
# also the case of both being None
385
elif old_entry is None:
316
return 'modified/renamed/reparented'
387
elif new_entry is None:
389
if old_entry.kind != new_entry.kind:
391
text_modified, meta_modified = new_entry.detect_changes(old_entry)
392
if text_modified or meta_modified:
396
# TODO 20060511 (mbp, rbc) factor out 'detect_rename' here.
397
if old_entry.parent_id != new_entry.parent_id:
399
elif old_entry.name != new_entry.name:
403
if renamed and not modified:
404
return InventoryEntry.RENAMED
405
if modified and not renamed:
407
if modified and renamed:
408
return InventoryEntry.MODIFIED_AND_RENAMED
318
411
def __repr__(self):
319
return ("%s(%r, %r, parent_id=%r)"
412
return ("%s(%r, %r, parent_id=%r, revision=%r)"
320
413
% (self.__class__.__name__,
325
419
def snapshot(self, revision, path, previous_entries,
326
work_tree, weave_store, transaction):
420
work_tree, commit_builder):
327
421
"""Make a snapshot of this entry which may or may not have changed.
329
423
This means that all its fields are populated, that it has its
330
424
text stored in the text store or weave.
332
mutter('new parents of %s are %r', path, previous_entries)
426
# mutter('new parents of %s are %r', path, previous_entries)
333
427
self._read_tree_state(path, work_tree)
428
# TODO: Where should we determine whether to reuse a
429
# previous revision id or create a new revision? 20060606
334
430
if len(previous_entries) == 1:
335
431
# cannot be unchanged unless there is only one parent file rev.
336
432
parent_ie = previous_entries.values()[0]
337
433
if self._unchanged(parent_ie):
338
mutter("found unchanged entry")
434
# mutter("found unchanged entry")
339
435
self.revision = parent_ie.revision
340
436
return "unchanged"
341
return self.snapshot_revision(revision, previous_entries,
342
work_tree, weave_store, transaction)
344
def snapshot_revision(self, revision, previous_entries, work_tree,
345
weave_store, transaction):
346
"""Record this revision unconditionally."""
347
mutter('new revision for {%s}', self.file_id)
437
return self._snapshot_into_revision(revision, previous_entries,
438
work_tree, commit_builder)
440
def _snapshot_into_revision(self, revision, previous_entries, work_tree,
442
"""Record this revision unconditionally into a store.
444
The entry's last-changed revision property (`revision`) is updated to
445
that of the new revision.
447
:param revision: id of the new revision that is being recorded.
449
:returns: String description of the commit (e.g. "merged", "modified"), etc.
451
# mutter('new revision {%s} for {%s}', revision, self.file_id)
348
452
self.revision = revision
349
change = self._get_snapshot_change(previous_entries)
350
self._snapshot_text(previous_entries, work_tree, weave_store,
453
self._snapshot_text(previous_entries, work_tree, commit_builder)
354
def _snapshot_text(self, file_parents, work_tree, weave_store, transaction):
455
def _snapshot_text(self, file_parents, work_tree, commit_builder):
355
456
"""Record the 'text' of this entry, whatever form that takes.
357
458
This default implementation simply adds an empty text.
359
mutter('storing file {%s} in revision {%s}',
360
self.file_id, self.revision)
361
self._add_text_to_weave([], file_parents, weave_store, transaction)
460
raise NotImplementedError(self._snapshot_text)
363
462
def __eq__(self, other):
364
463
if not isinstance(other, InventoryEntry):
407
509
# first requested, or preload them if they're already known
408
510
pass # nothing to do by default
512
def _forget_tree_state(self):
411
516
class RootEntry(InventoryEntry):
518
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
519
'text_id', 'parent_id', 'children', 'executable',
520
'revision', 'symlink_target', 'reference_revision']
413
522
def _check(self, checker, rev_id, tree):
414
523
"""See InventoryEntry._check"""
416
525
def __init__(self, file_id):
417
526
self.file_id = file_id
418
527
self.children = {}
419
self.kind = 'root_directory'
528
self.kind = 'directory'
420
529
self.parent_id = None
532
symbol_versioning.warn('RootEntry is deprecated as of bzr 0.10.'
533
' Please use InventoryDirectory instead.',
534
DeprecationWarning, stacklevel=2)
423
536
def __eq__(self, other):
424
537
if not isinstance(other, RootEntry):
431
544
class InventoryDirectory(InventoryEntry):
432
545
"""A directory in an inventory."""
547
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
548
'text_id', 'parent_id', 'children', 'executable',
549
'revision', 'symlink_target', 'reference_revision']
434
551
def _check(self, checker, rev_id, tree):
435
552
"""See InventoryEntry._check"""
436
if self.text_sha1 != None or self.text_size != None or self.text_id != None:
553
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
437
554
raise BzrCheckError('directory {%s} has text in revision {%s}'
438
555
% (self.file_id, rev_id))
466
583
"""See InventoryEntry._put_on_disk."""
467
584
os.mkdir(fullpath)
586
def _snapshot_text(self, file_parents, work_tree, commit_builder):
587
"""See InventoryEntry._snapshot_text."""
588
commit_builder.modified_directory(self.file_id, file_parents)
470
591
class InventoryFile(InventoryEntry):
471
592
"""A file in an inventory."""
473
def _check(self, checker, rev_id, tree):
594
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
595
'text_id', 'parent_id', 'children', 'executable',
596
'revision', 'symlink_target', 'reference_revision']
598
def _check(self, checker, tree_revision_id, tree):
474
599
"""See InventoryEntry._check"""
475
revision = self.revision
476
t = (self.file_id, revision)
600
t = (self.file_id, self.revision)
477
601
if t in checker.checked_texts:
478
prev_sha = checker.checked_texts[t]
602
prev_sha = checker.checked_texts[t]
479
603
if prev_sha != self.text_sha1:
480
604
raise BzrCheckError('mismatched sha1 on {%s} in {%s}' %
481
(self.file_id, rev_id))
605
(self.file_id, tree_revision_id))
483
607
checker.repeated_text_cnt += 1
485
mutter('check version {%s} of {%s}', rev_id, self.file_id)
486
file_lines = tree.get_file_lines(self.file_id)
487
checker.checked_text_cnt += 1
488
if self.text_size != sum(map(len, file_lines)):
489
raise BzrCheckError('text {%s} wrong size' % self.text_id)
490
if self.text_sha1 != sha_strings(file_lines):
491
raise BzrCheckError('text {%s} wrong sha1' % self.text_id)
610
if self.file_id not in checker.checked_weaves:
611
mutter('check weave {%s}', self.file_id)
612
w = tree.get_weave(self.file_id)
613
# Not passing a progress bar, because it creates a new
614
# progress, which overwrites the current progress,
615
# and doesn't look nice
617
checker.checked_weaves[self.file_id] = True
619
w = tree.get_weave(self.file_id)
621
mutter('check version {%s} of {%s}', tree_revision_id, self.file_id)
622
checker.checked_text_cnt += 1
623
# We can't check the length, because Weave doesn't store that
624
# information, and the whole point of looking at the weave's
625
# sha1sum is that we don't have to extract the text.
626
if self.text_sha1 != w.get_sha1(self.revision):
627
raise BzrCheckError('text {%s} version {%s} wrong sha1'
628
% (self.file_id, self.revision))
492
629
checker.checked_texts[t] = self.text_sha1
511
648
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
512
649
output_to, reverse=False):
513
650
"""See InventoryEntry._diff."""
514
from_text = tree.get_file(self.file_id).readlines()
516
to_text = to_tree.get_file(to_entry.file_id).readlines()
520
text_diff(from_label, from_text,
521
to_label, to_text, output_to)
523
text_diff(to_label, to_text,
524
from_label, from_text, output_to)
652
from_text = tree.get_file(self.file_id).readlines()
654
to_text = to_tree.get_file(to_entry.file_id).readlines()
658
text_diff(from_label, from_text,
659
to_label, to_text, output_to)
661
text_diff(to_label, to_text,
662
from_label, from_text, output_to)
663
except errors.BinaryFile:
665
label_pair = (to_label, from_label)
667
label_pair = (from_label, to_label)
668
print >> output_to, "Binary files %s and %s differ" % label_pair
526
670
def has_text(self):
527
671
"""See InventoryEntry.has_text."""
549
693
def _put_on_disk(self, fullpath, tree):
550
694
"""See InventoryEntry._put_on_disk."""
551
pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
695
osutils.pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
552
696
if tree.is_executable(self.file_id):
553
697
os.chmod(fullpath, 0755)
555
699
def _read_tree_state(self, path, work_tree):
556
700
"""See InventoryEntry._read_tree_state."""
557
self.text_sha1 = work_tree.get_file_sha1(self.file_id)
558
self.executable = work_tree.is_executable(self.file_id)
560
def _snapshot_text(self, file_parents, work_tree, weave_store, transaction):
701
self.text_sha1 = work_tree.get_file_sha1(self.file_id, path=path)
702
# FIXME: 20050930 probe for the text size when getting sha1
703
# in _read_tree_state
704
self.executable = work_tree.is_executable(self.file_id, path=path)
707
return ("%s(%r, %r, parent_id=%r, sha1=%r, len=%s)"
708
% (self.__class__.__name__,
715
def _forget_tree_state(self):
716
self.text_sha1 = None
718
def _snapshot_text(self, file_parents, work_tree, commit_builder):
561
719
"""See InventoryEntry._snapshot_text."""
562
mutter('storing file {%s} in revision {%s}',
563
self.file_id, self.revision)
564
# special case to avoid diffing on renames or
566
if (len(file_parents) == 1
567
and self.text_sha1 == file_parents.values()[0].text_sha1
568
and self.text_size == file_parents.values()[0].text_size):
569
previous_ie = file_parents.values()[0]
570
weave_store.add_identical_text(
571
self.file_id, previous_ie.revision,
572
self.revision, file_parents, transaction)
574
new_lines = work_tree.get_file(self.file_id).readlines()
575
self._add_text_to_weave(new_lines, file_parents, weave_store,
577
self.text_sha1 = sha_strings(new_lines)
578
self.text_size = sum(map(len, new_lines))
720
def get_content_byte_lines():
721
return work_tree.get_file(self.file_id).readlines()
722
self.text_sha1, self.text_size = commit_builder.modified_file_text(
723
self.file_id, file_parents, get_content_byte_lines, self.text_sha1, self.text_size)
581
725
def _unchanged(self, previous_ie):
582
726
"""See InventoryEntry._unchanged."""
595
739
class InventoryLink(InventoryEntry):
596
740
"""A file in an inventory."""
598
__slots__ = ['symlink_target']
742
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
743
'text_id', 'parent_id', 'children', 'executable',
744
'revision', 'symlink_target', 'reference_revision']
600
746
def _check(self, checker, rev_id, tree):
601
747
"""See InventoryEntry._check"""
602
if self.text_sha1 != None or self.text_size != None or self.text_id != None:
748
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
603
749
raise BzrCheckError('symlink {%s} has text in revision {%s}'
604
750
% (self.file_id, rev_id))
605
if self.symlink_target == None:
751
if self.symlink_target is None:
606
752
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
607
753
% (self.file_id, rev_id))
673
822
compatible = False
674
823
return compatible
825
def _snapshot_text(self, file_parents, work_tree, commit_builder):
826
"""See InventoryEntry._snapshot_text."""
827
commit_builder.modified_link(
828
self.file_id, file_parents, self.symlink_target)
831
class TreeReference(InventoryEntry):
833
kind = 'tree-reference'
835
def __init__(self, file_id, name, parent_id, revision=None,
836
reference_revision=None):
837
InventoryEntry.__init__(self, file_id, name, parent_id)
838
self.revision = revision
839
self.reference_revision = reference_revision
842
return TreeReference(self.file_id, self.name, self.parent_id,
843
self.revision, self.reference_revision)
845
def _snapshot_text(self, file_parents, work_tree, commit_builder):
846
commit_builder.modified_reference(self.file_id, file_parents)
848
def _read_tree_state(self, path, work_tree):
849
"""Populate fields in the inventory entry from the given tree.
851
self.reference_revision = work_tree.get_reference_revision(
854
def _forget_tree_state(self):
855
self.reference_revision = None
677
858
class Inventory(object):
678
859
"""Inventory of versioned files in a tree.
707
888
May also look up by name:
709
890
>>> [x[0] for x in inv.iter_entries()]
711
892
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
712
893
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
713
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT-12345678-12345678')
894
Traceback (most recent call last):
895
BzrError: parent_id {TREE_ROOT} not in inventory
896
>>> inv.add(InventoryFile('123-123', 'hello.c', 'TREE_ROOT-12345678-12345678'))
897
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT-12345678-12345678', sha1=None, len=None)
715
def __init__(self, root_id=ROOT_ID):
899
def __init__(self, root_id=ROOT_ID, revision_id=None):
716
900
"""Create or read an inventory.
718
902
If a working directory is specified, the inventory is read
722
906
The inventory is created with a default root directory, with
725
# We are letting Branch.initialize() create a unique inventory
726
# root id. Rather than generating a random one here.
728
# root_id = bzrlib.branch.gen_file_id('TREE_ROOT')
729
self.root = RootEntry(root_id)
909
if root_id is not None:
910
assert root_id.__class__ == str
911
self._set_root(InventoryDirectory(root_id, u'', None))
915
self.revision_id = revision_id
917
def _set_root(self, ie):
730
919
self._byid = {self.root.file_id: self.root}
734
other = Inventory(self.root.file_id)
922
# TODO: jam 20051218 Should copy also copy the revision_id?
923
entries = self.iter_entries()
924
other = Inventory(entries.next()[1].file_id)
735
925
# copy recursively so we know directories will be added before
736
926
# their children. There are more efficient ways than this...
737
for path, entry in self.iter_entries():
738
if entry == self.root:
927
for path, entry in entries():
740
928
other.add(entry.copy())
744
931
def __iter__(self):
745
932
return iter(self._byid)
748
934
def __len__(self):
749
935
"""Returns number of entries."""
750
936
return len(self._byid)
753
938
def iter_entries(self, from_dir=None):
754
939
"""Return (path, entry) pairs, in order by name."""
758
elif isinstance(from_dir, basestring):
759
from_dir = self._byid[from_dir]
761
kids = from_dir.children.items()
763
for name, ie in kids:
765
if ie.kind == 'directory':
766
for cn, cie in self.iter_entries(from_dir=ie.file_id):
767
yield os.path.join(name, cn), cie
941
if self.root is None:
945
elif isinstance(from_dir, basestring):
946
from_dir = self._byid[from_dir]
948
# unrolling the recursive called changed the time from
949
# 440ms/663ms (inline/total) to 116ms/116ms
950
children = from_dir.children.items()
952
children = collections.deque(children)
953
stack = [(u'', children)]
955
from_dir_relpath, children = stack[-1]
958
name, ie = children.popleft()
960
# we know that from_dir_relpath never ends in a slash
961
# and 'f' doesn't begin with one, we can do a string op, rather
962
# than the checks of pathjoin(), though this means that all paths
964
path = from_dir_relpath + '/' + name
968
if ie.kind != 'directory':
971
# But do this child first
972
new_children = ie.children.items()
974
new_children = collections.deque(new_children)
975
stack.append((path, new_children))
976
# Break out of inner loop, so that we start outer loop with child
979
# if we finished all children, pop it off the stack
982
def iter_entries_by_dir(self, from_dir=None, specific_file_ids=None):
983
"""Iterate over the entries in a directory first order.
985
This returns all entries for a directory before returning
986
the entries for children of a directory. This is not
987
lexicographically sorted order, and is a hybrid between
988
depth-first and breadth-first.
990
:return: This yields (path, entry) pairs
992
if specific_file_ids:
993
safe = osutils.safe_file_id
994
specific_file_ids = set(safe(fid) for fid in specific_file_ids)
995
# TODO? Perhaps this should return the from_dir so that the root is
996
# yielded? or maybe an option?
998
if self.root is None:
1000
# Optimize a common case
1001
if specific_file_ids is not None and len(specific_file_ids) == 1:
1002
file_id = list(specific_file_ids)[0]
1004
yield self.id2path(file_id), self[file_id]
1006
from_dir = self.root
1007
if (specific_file_ids is None or
1008
self.root.file_id in specific_file_ids):
1009
yield u'', self.root
1010
elif isinstance(from_dir, basestring):
1011
from_dir = self._byid[from_dir]
1013
if specific_file_ids is not None:
1014
# TODO: jam 20070302 This could really be done as a loop rather
1015
# than a bunch of recursive calls.
1018
def add_ancestors(file_id):
1019
if file_id not in byid:
1021
parent_id = byid[file_id].parent_id
1022
if parent_id is None:
1024
if parent_id not in parents:
1025
parents.add(parent_id)
1026
add_ancestors(parent_id)
1027
for file_id in specific_file_ids:
1028
add_ancestors(file_id)
1032
stack = [(u'', from_dir)]
1034
cur_relpath, cur_dir = stack.pop()
1037
for child_name, child_ie in sorted(cur_dir.children.iteritems()):
1039
child_relpath = cur_relpath + child_name
1041
if (specific_file_ids is None or
1042
child_ie.file_id in specific_file_ids):
1043
yield child_relpath, child_ie
1045
if child_ie.kind == 'directory':
1046
if parents is None or child_ie.file_id in parents:
1047
child_dirs.append((child_relpath+'/', child_ie))
1048
stack.extend(reversed(child_dirs))
770
1050
def entries(self):
771
1051
"""Return list of (path, ie) for all entries except the root.
799
1078
for name, child_ie in kids:
800
child_path = os.path.join(parent_path, name)
1079
child_path = osutils.pathjoin(parent_path, name)
801
1080
descend(child_ie, child_path)
802
descend(self.root, '')
1081
descend(self.root, u'')
807
1084
def __contains__(self, file_id):
808
1085
"""True if this entry contains a file with given id.
810
1087
>>> inv = Inventory()
811
1088
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
812
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT')
1089
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
813
1090
>>> '123' in inv
815
1092
>>> '456' in inv
818
return file_id in self._byid
1095
file_id = osutils.safe_file_id(file_id)
1096
return (file_id in self._byid)
821
1098
def __getitem__(self, file_id):
822
1099
"""Return the entry for given file_id.
824
1101
>>> inv = Inventory()
825
1102
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
826
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT')
1103
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
827
1104
>>> inv['123123'].name
1107
file_id = osutils.safe_file_id(file_id)
831
1109
return self._byid[file_id]
832
1110
except KeyError:
834
raise BzrError("can't look up file_id None")
836
raise BzrError("file_id {%s} not in inventory" % file_id)
1111
# really we're passing an inventory, not a tree...
1112
raise errors.NoSuchId(self, file_id)
839
1114
def get_file_kind(self, file_id):
1115
file_id = osutils.safe_file_id(file_id)
840
1116
return self._byid[file_id].kind
842
1118
def get_child(self, parent_id, filename):
1119
parent_id = osutils.safe_file_id(parent_id)
843
1120
return self[parent_id].children.get(filename)
1122
def _add_child(self, entry):
1123
"""Add an entry to the inventory, without adding it to its parent"""
1124
if entry.file_id in self._byid:
1125
raise BzrError("inventory already contains entry with id {%s}" %
1127
self._byid[entry.file_id] = entry
1128
for child in getattr(entry, 'children', {}).itervalues():
1129
self._add_child(child)
846
1132
def add(self, entry):
847
1133
"""Add entry to inventory.
852
1138
Returns the new entry object.
854
1140
if entry.file_id in self._byid:
855
raise BzrError("inventory already contains entry with id {%s}" % entry.file_id)
857
if entry.parent_id == ROOT_ID or entry.parent_id is None:
858
entry.parent_id = self.root.file_id
861
parent = self._byid[entry.parent_id]
863
raise BzrError("parent_id {%s} not in inventory" % entry.parent_id)
865
if parent.children.has_key(entry.name):
866
raise BzrError("%s is already versioned" %
867
appendpath(self.id2path(parent.file_id), entry.name))
869
self._byid[entry.file_id] = entry
870
parent.children[entry.name] = entry
874
def add_path(self, relpath, kind, file_id=None):
1141
raise errors.DuplicateFileId(entry.file_id,
1142
self._byid[entry.file_id])
1144
if entry.parent_id is None:
1145
assert self.root is None and len(self._byid) == 0
1149
parent = self._byid[entry.parent_id]
1151
raise BzrError("parent_id {%s} not in inventory" %
1154
if entry.name in parent.children:
1155
raise BzrError("%s is already versioned" %
1156
osutils.pathjoin(self.id2path(parent.file_id),
1157
entry.name).encode('utf-8'))
1158
parent.children[entry.name] = entry
1159
return self._add_child(entry)
1161
def add_path(self, relpath, kind, file_id=None, parent_id=None):
875
1162
"""Add entry from a path.
877
1164
The immediate parent must already be versioned.
879
1166
Returns the new entry object."""
880
from bzrlib.branch import gen_file_id
882
parts = bzrlib.osutils.splitpath(relpath)
1168
parts = osutils.splitpath(relpath)
883
1170
if len(parts) == 0:
884
raise BzrError("cannot re-add root of inventory")
887
file_id = gen_file_id(relpath)
889
parent_path = parts[:-1]
890
parent_id = self.path2id(parent_path)
891
if parent_id == None:
892
raise NotVersionedError(path=parent_path)
893
if kind == 'directory':
894
ie = InventoryDirectory(file_id, parts[-1], parent_id)
896
ie = InventoryFile(file_id, parts[-1], parent_id)
897
elif kind == 'symlink':
898
ie = InventoryLink(file_id, parts[-1], parent_id)
1172
file_id = generate_ids.gen_root_id()
1174
file_id = osutils.safe_file_id(file_id)
1175
self.root = InventoryDirectory(file_id, '', None)
1176
self._byid = {self.root.file_id: self.root}
900
raise BzrError("unknown kind %r" % kind)
1179
parent_path = parts[:-1]
1180
parent_id = self.path2id(parent_path)
1181
if parent_id is None:
1182
raise errors.NotVersionedError(path=parent_path)
1183
ie = make_entry(kind, parts[-1], parent_id, file_id)
901
1184
return self.add(ie)
904
1186
def __delitem__(self, file_id):
905
1187
"""Remove entry by id.
907
1189
>>> inv = Inventory()
908
1190
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
909
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT')
1191
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
910
1192
>>> '123' in inv
912
1194
>>> del inv['123']
913
1195
>>> '123' in inv
1198
file_id = osutils.safe_file_id(file_id)
916
1199
ie = self[file_id]
918
assert self[ie.parent_id].children[ie.name] == ie
1201
assert ie.parent_id is None or \
1202
self[ie.parent_id].children[ie.name] == ie
920
# TODO: Test deleting all children; maybe hoist to a separate
922
if ie.kind == 'directory':
923
for cie in ie.children.values():
924
del self[cie.file_id]
927
1204
del self._byid[file_id]
928
del self[ie.parent_id].children[ie.name]
1205
if ie.parent_id is not None:
1206
del self[ie.parent_id].children[ie.name]
931
1208
def __eq__(self, other):
932
1209
"""Compare two sets by comparing their contents.
938
1215
>>> i1.add(InventoryFile('123', 'foo', ROOT_ID))
939
InventoryFile('123', 'foo', parent_id='TREE_ROOT')
1216
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
942
1219
>>> i2.add(InventoryFile('123', 'foo', ROOT_ID))
943
InventoryFile('123', 'foo', parent_id='TREE_ROOT')
1220
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
947
1224
if not isinstance(other, Inventory):
948
1225
return NotImplemented
950
if len(self._byid) != len(other._byid):
951
# shortcut: obviously not the same
954
1227
return self._byid == other._byid
957
1229
def __ne__(self, other):
958
1230
return not self.__eq__(other)
961
1232
def __hash__(self):
962
1233
raise ValueError('not hashable')
1235
def _iter_file_id_parents(self, file_id):
1236
"""Yield the parents of file_id up to the root."""
1237
file_id = osutils.safe_file_id(file_id)
1238
while file_id is not None:
1240
ie = self._byid[file_id]
1242
raise errors.NoSuchId(tree=None, file_id=file_id)
1244
file_id = ie.parent_id
965
1246
def get_idpath(self, file_id):
966
1247
"""Return a list of file_ids for the path to an entry.
970
1251
is equal to the depth of the file in the tree, counting the
971
1252
root directory as depth 1.
1254
file_id = osutils.safe_file_id(file_id)
974
while file_id != None:
976
ie = self._byid[file_id]
978
raise BzrError("file_id {%s} not found in inventory" % file_id)
979
p.insert(0, ie.file_id)
980
file_id = ie.parent_id
1256
for parent in self._iter_file_id_parents(file_id):
1257
p.insert(0, parent.file_id)
984
1260
def id2path(self, file_id):
985
"""Return as a list the path to file_id.
1261
"""Return as a string the path to file_id.
987
1263
>>> i = Inventory()
988
1264
>>> e = i.add(InventoryDirectory('src-id', 'src', ROOT_ID))
989
1265
>>> e = i.add(InventoryFile('foo-id', 'foo.c', parent_id='src-id'))
990
>>> print i.id2path('foo-id').replace(os.sep, '/')
1266
>>> print i.id2path('foo-id')
1269
file_id = osutils.safe_file_id(file_id)
993
1270
# get all names, skipping root
994
p = [self._byid[fid].name for fid in self.get_idpath(file_id)[1:]]
995
return os.sep.join(p)
1271
return '/'.join(reversed(
1272
[parent.name for parent in
1273
self._iter_file_id_parents(file_id)][:-1]))
999
1275
def path2id(self, name):
1000
1276
"""Walk down through directories to return entry of last component.
1026
1307
return parent.file_id
1029
1309
def has_filename(self, names):
1030
1310
return bool(self.path2id(names))
1033
1312
def has_id(self, file_id):
1034
return self._byid.has_key(file_id)
1313
file_id = osutils.safe_file_id(file_id)
1314
return (file_id in self._byid)
1316
def remove_recursive_id(self, file_id):
1317
"""Remove file_id, and children, from the inventory.
1319
:param file_id: A file_id to remove.
1321
file_id = osutils.safe_file_id(file_id)
1322
to_find_delete = [self._byid[file_id]]
1324
while to_find_delete:
1325
ie = to_find_delete.pop()
1326
to_delete.append(ie.file_id)
1327
if ie.kind == 'directory':
1328
to_find_delete.extend(ie.children.values())
1329
for file_id in reversed(to_delete):
1331
del self._byid[file_id]
1332
if ie.parent_id is not None:
1333
del self[ie.parent_id].children[ie.name]
1037
1337
def rename(self, file_id, new_parent_id, new_name):
1038
1338
"""Move a file within the inventory.
1040
1340
This can change either the name, or the parent, or both.
1042
This does not move the working file."""
1342
This does not move the working file.
1344
file_id = osutils.safe_file_id(file_id)
1043
1345
if not is_valid_name(new_name):
1044
1346
raise BzrError("not an acceptable filename: %r" % new_name)
1063
1365
file_ie.name = new_name
1064
1366
file_ie.parent_id = new_parent_id
1368
def is_root(self, file_id):
1369
file_id = osutils.safe_file_id(file_id)
1370
return self.root is not None and file_id == self.root.file_id
1374
'directory': InventoryDirectory,
1375
'file': InventoryFile,
1376
'symlink': InventoryLink,
1377
'tree-reference': TreeReference
1380
def make_entry(kind, name, parent_id, file_id=None):
1381
"""Create an inventory entry.
1383
:param kind: the type of inventory entry to create.
1384
:param name: the basename of the entry.
1385
:param parent_id: the parent_id of the entry.
1386
:param file_id: the file_id to use. if None, one will be created.
1389
file_id = generate_ids.gen_file_id(name)
1391
file_id = osutils.safe_file_id(file_id)
1393
#------- This has been copied to bzrlib.dirstate.DirState.add, please
1394
# keep them synchronised.
1395
# we dont import normalized_filename directly because we want to be
1396
# able to change the implementation at runtime for tests.
1397
norm_name, can_access = osutils.normalized_filename(name)
1398
if norm_name != name:
1402
# TODO: jam 20060701 This would probably be more useful
1403
# if the error was raised with the full path
1404
raise errors.InvalidNormalization(name)
1407
factory = entry_factory[kind]
1409
raise BzrError("unknown kind %r" % kind)
1410
return factory(file_id, name, parent_id)
1069
1413
_NAME_RE = None
1071
1415
def is_valid_name(name):
1072
1416
global _NAME_RE
1073
if _NAME_RE == None:
1417
if _NAME_RE is None:
1074
1418
_NAME_RE = re.compile(r'^[^/\\]+$')
1076
1420
return bool(_NAME_RE.match(name))
added
removed
bzrlib/inventory.py
