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: pathjoin('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():
106
114
... assert i.path2id(path)
113
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, topo_sorted=False))
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(pathjoin(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 = pathjoin(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
486
610
if self.file_id not in checker.checked_weaves:
487
611
mutter('check weave {%s}', self.file_id)
488
w = tree.get_weave(self.file_id)
612
w = tree._get_weave(self.file_id)
489
613
# Not passing a progress bar, because it creates a new
490
614
# progress, which overwrites the current progress,
491
615
# and doesn't look nice
493
617
checker.checked_weaves[self.file_id] = True
495
w = tree.get_weave_prelude(self.file_id)
619
w = tree._get_weave(self.file_id)
497
mutter('check version {%s} of {%s}', rev_id, self.file_id)
498
checker.checked_text_cnt += 1
621
mutter('check version {%s} of {%s}', tree_revision_id, self.file_id)
622
checker.checked_text_cnt += 1
499
623
# We can't check the length, because Weave doesn't store that
500
624
# information, and the whole point of looking at the weave's
501
625
# sha1sum is that we don't have to extract the text.
524
648
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
525
649
output_to, reverse=False):
526
650
"""See InventoryEntry._diff."""
527
from_text = tree.get_file(self.file_id).readlines()
529
to_text = to_tree.get_file(to_entry.file_id).readlines()
533
text_diff(from_label, from_text,
534
to_label, to_text, output_to)
536
text_diff(to_label, to_text,
537
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, \
669
("Binary files %s and %s differ" % label_pair).encode('utf8')
539
671
def has_text(self):
540
672
"""See InventoryEntry.has_text."""
562
694
def _put_on_disk(self, fullpath, tree):
563
695
"""See InventoryEntry._put_on_disk."""
564
pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
696
osutils.pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
565
697
if tree.is_executable(self.file_id):
566
698
os.chmod(fullpath, 0755)
568
700
def _read_tree_state(self, path, work_tree):
569
701
"""See InventoryEntry._read_tree_state."""
570
self.text_sha1 = work_tree.get_file_sha1(self.file_id)
571
self.executable = work_tree.is_executable(self.file_id)
573
def _snapshot_text(self, file_parents, work_tree, weave_store, transaction):
702
self.text_sha1 = work_tree.get_file_sha1(self.file_id, path=path)
703
# FIXME: 20050930 probe for the text size when getting sha1
704
# in _read_tree_state
705
self.executable = work_tree.is_executable(self.file_id, path=path)
708
return ("%s(%r, %r, parent_id=%r, sha1=%r, len=%s)"
709
% (self.__class__.__name__,
716
def _forget_tree_state(self):
717
self.text_sha1 = None
719
def _snapshot_text(self, file_parents, work_tree, commit_builder):
574
720
"""See InventoryEntry._snapshot_text."""
575
mutter('storing file {%s} in revision {%s}',
576
self.file_id, self.revision)
577
# special case to avoid diffing on renames or
579
if (len(file_parents) == 1
580
and self.text_sha1 == file_parents.values()[0].text_sha1
581
and self.text_size == file_parents.values()[0].text_size):
582
previous_ie = file_parents.values()[0]
583
weave_store.add_identical_text(
584
self.file_id, previous_ie.revision,
585
self.revision, file_parents, transaction)
587
new_lines = work_tree.get_file(self.file_id).readlines()
588
self._add_text_to_weave(new_lines, file_parents, weave_store,
590
self.text_sha1 = sha_strings(new_lines)
591
self.text_size = sum(map(len, new_lines))
721
def get_content_byte_lines():
722
return work_tree.get_file(self.file_id).readlines()
723
self.text_sha1, self.text_size = commit_builder.modified_file_text(
724
self.file_id, file_parents, get_content_byte_lines, self.text_sha1, self.text_size)
594
726
def _unchanged(self, previous_ie):
595
727
"""See InventoryEntry._unchanged."""
608
740
class InventoryLink(InventoryEntry):
609
741
"""A file in an inventory."""
611
__slots__ = ['symlink_target']
743
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
744
'text_id', 'parent_id', 'children', 'executable',
745
'revision', 'symlink_target', 'reference_revision']
613
747
def _check(self, checker, rev_id, tree):
614
748
"""See InventoryEntry._check"""
615
if self.text_sha1 != None or self.text_size != None or self.text_id != None:
749
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
616
750
raise BzrCheckError('symlink {%s} has text in revision {%s}'
617
751
% (self.file_id, rev_id))
618
if self.symlink_target == None:
752
if self.symlink_target is None:
619
753
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
620
754
% (self.file_id, rev_id))
686
823
compatible = False
687
824
return compatible
826
def _snapshot_text(self, file_parents, work_tree, commit_builder):
827
"""See InventoryEntry._snapshot_text."""
828
commit_builder.modified_link(
829
self.file_id, file_parents, self.symlink_target)
832
class TreeReference(InventoryEntry):
834
kind = 'tree-reference'
836
def __init__(self, file_id, name, parent_id, revision=None,
837
reference_revision=None):
838
InventoryEntry.__init__(self, file_id, name, parent_id)
839
self.revision = revision
840
self.reference_revision = reference_revision
843
return TreeReference(self.file_id, self.name, self.parent_id,
844
self.revision, self.reference_revision)
846
def _snapshot_text(self, file_parents, work_tree, commit_builder):
847
commit_builder.modified_reference(self.file_id, file_parents)
849
def _read_tree_state(self, path, work_tree):
850
"""Populate fields in the inventory entry from the given tree.
852
self.reference_revision = work_tree.get_reference_revision(
855
def _forget_tree_state(self):
856
self.reference_revision = None
690
859
class Inventory(object):
691
860
"""Inventory of versioned files in a tree.
720
889
May also look up by name:
722
891
>>> [x[0] for x in inv.iter_entries()]
724
893
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
725
894
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
726
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT-12345678-12345678')
895
Traceback (most recent call last):
896
BzrError: parent_id {TREE_ROOT} not in inventory
897
>>> inv.add(InventoryFile('123-123', 'hello.c', 'TREE_ROOT-12345678-12345678'))
898
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT-12345678-12345678', sha1=None, len=None)
728
def __init__(self, root_id=ROOT_ID):
900
def __init__(self, root_id=ROOT_ID, revision_id=None):
729
901
"""Create or read an inventory.
731
903
If a working directory is specified, the inventory is read
735
907
The inventory is created with a default root directory, with
738
# We are letting Branch.initialize() create a unique inventory
739
# root id. Rather than generating a random one here.
741
# root_id = bzrlib.branch.gen_file_id('TREE_ROOT')
742
self.root = RootEntry(root_id)
910
if root_id is not None:
911
assert root_id.__class__ == str
912
self._set_root(InventoryDirectory(root_id, u'', None))
916
self.revision_id = revision_id
918
def _set_root(self, ie):
743
920
self._byid = {self.root.file_id: self.root}
747
other = Inventory(self.root.file_id)
923
# TODO: jam 20051218 Should copy also copy the revision_id?
924
entries = self.iter_entries()
925
other = Inventory(entries.next()[1].file_id)
748
926
# copy recursively so we know directories will be added before
749
927
# their children. There are more efficient ways than this...
750
for path, entry in self.iter_entries():
751
if entry == self.root:
928
for path, entry in entries():
753
929
other.add(entry.copy())
757
932
def __iter__(self):
758
933
return iter(self._byid)
761
935
def __len__(self):
762
936
"""Returns number of entries."""
763
937
return len(self._byid)
766
939
def iter_entries(self, from_dir=None):
767
940
"""Return (path, entry) pairs, in order by name."""
771
elif isinstance(from_dir, basestring):
772
from_dir = self._byid[from_dir]
774
kids = from_dir.children.items()
776
for name, ie in kids:
778
if ie.kind == 'directory':
779
for cn, cie in self.iter_entries(from_dir=ie.file_id):
780
yield pathjoin(name, cn), cie
942
if self.root is None:
946
elif isinstance(from_dir, basestring):
947
from_dir = self._byid[from_dir]
949
# unrolling the recursive called changed the time from
950
# 440ms/663ms (inline/total) to 116ms/116ms
951
children = from_dir.children.items()
953
children = collections.deque(children)
954
stack = [(u'', children)]
956
from_dir_relpath, children = stack[-1]
959
name, ie = children.popleft()
961
# we know that from_dir_relpath never ends in a slash
962
# and 'f' doesn't begin with one, we can do a string op, rather
963
# than the checks of pathjoin(), though this means that all paths
965
path = from_dir_relpath + '/' + name
969
if ie.kind != 'directory':
972
# But do this child first
973
new_children = ie.children.items()
975
new_children = collections.deque(new_children)
976
stack.append((path, new_children))
977
# Break out of inner loop, so that we start outer loop with child
980
# if we finished all children, pop it off the stack
983
def iter_entries_by_dir(self, from_dir=None, specific_file_ids=None):
984
"""Iterate over the entries in a directory first order.
986
This returns all entries for a directory before returning
987
the entries for children of a directory. This is not
988
lexicographically sorted order, and is a hybrid between
989
depth-first and breadth-first.
991
:return: This yields (path, entry) pairs
993
if specific_file_ids:
994
safe = osutils.safe_file_id
995
specific_file_ids = set(safe(fid) for fid in specific_file_ids)
996
# TODO? Perhaps this should return the from_dir so that the root is
997
# yielded? or maybe an option?
999
if self.root is None:
1001
# Optimize a common case
1002
if specific_file_ids is not None and len(specific_file_ids) == 1:
1003
file_id = list(specific_file_ids)[0]
1005
yield self.id2path(file_id), self[file_id]
1007
from_dir = self.root
1008
if (specific_file_ids is None or
1009
self.root.file_id in specific_file_ids):
1010
yield u'', self.root
1011
elif isinstance(from_dir, basestring):
1012
from_dir = self._byid[from_dir]
1014
if specific_file_ids is not None:
1015
# TODO: jam 20070302 This could really be done as a loop rather
1016
# than a bunch of recursive calls.
1019
def add_ancestors(file_id):
1020
if file_id not in byid:
1022
parent_id = byid[file_id].parent_id
1023
if parent_id is None:
1025
if parent_id not in parents:
1026
parents.add(parent_id)
1027
add_ancestors(parent_id)
1028
for file_id in specific_file_ids:
1029
add_ancestors(file_id)
1033
stack = [(u'', from_dir)]
1035
cur_relpath, cur_dir = stack.pop()
1038
for child_name, child_ie in sorted(cur_dir.children.iteritems()):
1040
child_relpath = cur_relpath + child_name
1042
if (specific_file_ids is None or
1043
child_ie.file_id in specific_file_ids):
1044
yield child_relpath, child_ie
1046
if child_ie.kind == 'directory':
1047
if parents is None or child_ie.file_id in parents:
1048
child_dirs.append((child_relpath+'/', child_ie))
1049
stack.extend(reversed(child_dirs))
1051
def make_entry(self, kind, name, parent_id, file_id=None):
1052
"""Simple thunk to bzrlib.inventory.make_entry."""
1053
return make_entry(kind, name, parent_id, file_id)
783
1055
def entries(self):
784
1056
"""Return list of (path, ie) for all entries except the root.
812
1083
for name, child_ie in kids:
813
child_path = pathjoin(parent_path, name)
1084
child_path = osutils.pathjoin(parent_path, name)
814
1085
descend(child_ie, child_path)
815
1086
descend(self.root, u'')
820
1089
def __contains__(self, file_id):
821
1090
"""True if this entry contains a file with given id.
823
1092
>>> inv = Inventory()
824
1093
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
825
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT')
1094
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
826
1095
>>> '123' in inv
828
1097
>>> '456' in inv
831
return file_id in self._byid
1100
file_id = osutils.safe_file_id(file_id)
1101
return (file_id in self._byid)
834
1103
def __getitem__(self, file_id):
835
1104
"""Return the entry for given file_id.
837
1106
>>> inv = Inventory()
838
1107
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
839
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT')
1108
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
840
1109
>>> inv['123123'].name
1112
file_id = osutils.safe_file_id(file_id)
844
1114
return self._byid[file_id]
845
1115
except KeyError:
847
raise BzrError("can't look up file_id None")
849
raise BzrError("file_id {%s} not in inventory" % file_id)
1116
# really we're passing an inventory, not a tree...
1117
raise errors.NoSuchId(self, file_id)
852
1119
def get_file_kind(self, file_id):
1120
file_id = osutils.safe_file_id(file_id)
853
1121
return self._byid[file_id].kind
855
1123
def get_child(self, parent_id, filename):
1124
parent_id = osutils.safe_file_id(parent_id)
856
1125
return self[parent_id].children.get(filename)
1127
def _add_child(self, entry):
1128
"""Add an entry to the inventory, without adding it to its parent"""
1129
if entry.file_id in self._byid:
1130
raise BzrError("inventory already contains entry with id {%s}" %
1132
self._byid[entry.file_id] = entry
1133
for child in getattr(entry, 'children', {}).itervalues():
1134
self._add_child(child)
859
1137
def add(self, entry):
860
1138
"""Add entry to inventory.
865
1143
Returns the new entry object.
867
1145
if entry.file_id in self._byid:
868
raise BzrError("inventory already contains entry with id {%s}" % entry.file_id)
870
if entry.parent_id == ROOT_ID or entry.parent_id is None:
871
entry.parent_id = self.root.file_id
874
parent = self._byid[entry.parent_id]
876
raise BzrError("parent_id {%s} not in inventory" % entry.parent_id)
878
if parent.children.has_key(entry.name):
879
raise BzrError("%s is already versioned" %
880
pathjoin(self.id2path(parent.file_id), entry.name))
882
self._byid[entry.file_id] = entry
883
parent.children[entry.name] = entry
887
def add_path(self, relpath, kind, file_id=None):
1146
raise errors.DuplicateFileId(entry.file_id,
1147
self._byid[entry.file_id])
1149
if entry.parent_id is None:
1150
assert self.root is None and len(self._byid) == 0
1154
parent = self._byid[entry.parent_id]
1156
raise BzrError("parent_id {%s} not in inventory" %
1159
if entry.name in parent.children:
1160
raise BzrError("%s is already versioned" %
1161
osutils.pathjoin(self.id2path(parent.file_id),
1162
entry.name).encode('utf-8'))
1163
parent.children[entry.name] = entry
1164
return self._add_child(entry)
1166
def add_path(self, relpath, kind, file_id=None, parent_id=None):
888
1167
"""Add entry from a path.
890
1169
The immediate parent must already be versioned.
892
1171
Returns the new entry object."""
893
from bzrlib.workingtree import gen_file_id
895
parts = bzrlib.osutils.splitpath(relpath)
1173
parts = osutils.splitpath(relpath)
896
1175
if len(parts) == 0:
897
raise BzrError("cannot re-add root of inventory")
900
file_id = gen_file_id(relpath)
902
parent_path = parts[:-1]
903
parent_id = self.path2id(parent_path)
904
if parent_id == None:
905
raise NotVersionedError(path=parent_path)
906
if kind == 'directory':
907
ie = InventoryDirectory(file_id, parts[-1], parent_id)
909
ie = InventoryFile(file_id, parts[-1], parent_id)
910
elif kind == 'symlink':
911
ie = InventoryLink(file_id, parts[-1], parent_id)
1177
file_id = generate_ids.gen_root_id()
1179
file_id = osutils.safe_file_id(file_id)
1180
self.root = InventoryDirectory(file_id, '', None)
1181
self._byid = {self.root.file_id: self.root}
913
raise BzrError("unknown kind %r" % kind)
1184
parent_path = parts[:-1]
1185
parent_id = self.path2id(parent_path)
1186
if parent_id is None:
1187
raise errors.NotVersionedError(path=parent_path)
1188
ie = make_entry(kind, parts[-1], parent_id, file_id)
914
1189
return self.add(ie)
917
1191
def __delitem__(self, file_id):
918
1192
"""Remove entry by id.
920
1194
>>> inv = Inventory()
921
1195
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
922
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT')
1196
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
923
1197
>>> '123' in inv
925
1199
>>> del inv['123']
926
1200
>>> '123' in inv
1203
file_id = osutils.safe_file_id(file_id)
929
1204
ie = self[file_id]
931
assert self[ie.parent_id].children[ie.name] == ie
1206
assert ie.parent_id is None or \
1207
self[ie.parent_id].children[ie.name] == ie
933
# TODO: Test deleting all children; maybe hoist to a separate
935
if ie.kind == 'directory':
936
for cie in ie.children.values():
937
del self[cie.file_id]
940
1209
del self._byid[file_id]
941
del self[ie.parent_id].children[ie.name]
1210
if ie.parent_id is not None:
1211
del self[ie.parent_id].children[ie.name]
944
1213
def __eq__(self, other):
945
1214
"""Compare two sets by comparing their contents.
951
1220
>>> i1.add(InventoryFile('123', 'foo', ROOT_ID))
952
InventoryFile('123', 'foo', parent_id='TREE_ROOT')
1221
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
955
1224
>>> i2.add(InventoryFile('123', 'foo', ROOT_ID))
956
InventoryFile('123', 'foo', parent_id='TREE_ROOT')
1225
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
960
1229
if not isinstance(other, Inventory):
961
1230
return NotImplemented
963
if len(self._byid) != len(other._byid):
964
# shortcut: obviously not the same
967
1232
return self._byid == other._byid
970
1234
def __ne__(self, other):
971
1235
return not self.__eq__(other)
974
1237
def __hash__(self):
975
1238
raise ValueError('not hashable')
1240
def _iter_file_id_parents(self, file_id):
1241
"""Yield the parents of file_id up to the root."""
1242
file_id = osutils.safe_file_id(file_id)
1243
while file_id is not None:
1245
ie = self._byid[file_id]
1247
raise errors.NoSuchId(tree=None, file_id=file_id)
1249
file_id = ie.parent_id
978
1251
def get_idpath(self, file_id):
979
1252
"""Return a list of file_ids for the path to an entry.
1042
1312
return parent.file_id
1045
1314
def has_filename(self, names):
1046
1315
return bool(self.path2id(names))
1049
1317
def has_id(self, file_id):
1050
return self._byid.has_key(file_id)
1318
file_id = osutils.safe_file_id(file_id)
1319
return (file_id in self._byid)
1321
def remove_recursive_id(self, file_id):
1322
"""Remove file_id, and children, from the inventory.
1324
:param file_id: A file_id to remove.
1326
file_id = osutils.safe_file_id(file_id)
1327
to_find_delete = [self._byid[file_id]]
1329
while to_find_delete:
1330
ie = to_find_delete.pop()
1331
to_delete.append(ie.file_id)
1332
if ie.kind == 'directory':
1333
to_find_delete.extend(ie.children.values())
1334
for file_id in reversed(to_delete):
1336
del self._byid[file_id]
1337
if ie.parent_id is not None:
1338
del self[ie.parent_id].children[ie.name]
1053
1342
def rename(self, file_id, new_parent_id, new_name):
1054
1343
"""Move a file within the inventory.
1056
1345
This can change either the name, or the parent, or both.
1058
This does not move the working file."""
1347
This does not move the working file.
1349
file_id = osutils.safe_file_id(file_id)
1059
1350
if not is_valid_name(new_name):
1060
1351
raise BzrError("not an acceptable filename: %r" % new_name)
1079
1370
file_ie.name = new_name
1080
1371
file_ie.parent_id = new_parent_id
1373
def is_root(self, file_id):
1374
file_id = osutils.safe_file_id(file_id)
1375
return self.root is not None and file_id == self.root.file_id
1379
'directory': InventoryDirectory,
1380
'file': InventoryFile,
1381
'symlink': InventoryLink,
1382
'tree-reference': TreeReference
1385
def make_entry(kind, name, parent_id, file_id=None):
1386
"""Create an inventory entry.
1388
:param kind: the type of inventory entry to create.
1389
:param name: the basename of the entry.
1390
:param parent_id: the parent_id of the entry.
1391
:param file_id: the file_id to use. if None, one will be created.
1394
file_id = generate_ids.gen_file_id(name)
1396
file_id = osutils.safe_file_id(file_id)
1398
#------- This has been copied to bzrlib.dirstate.DirState.add, please
1399
# keep them synchronised.
1400
# we dont import normalized_filename directly because we want to be
1401
# able to change the implementation at runtime for tests.
1402
norm_name, can_access = osutils.normalized_filename(name)
1403
if norm_name != name:
1407
# TODO: jam 20060701 This would probably be more useful
1408
# if the error was raised with the full path
1409
raise errors.InvalidNormalization(name)
1412
factory = entry_factory[kind]
1414
raise BzrError("unknown kind %r" % kind)
1415
return factory(file_id, name, parent_id)
1085
1418
_NAME_RE = None
1087
1420
def is_valid_name(name):
1088
1421
global _NAME_RE
1089
if _NAME_RE == None:
1422
if _NAME_RE is None:
1090
1423
_NAME_RE = re.compile(r'^[^/\\]+$')
1092
1425
return bool(_NAME_RE.match(name))
added
removed
bzrlib/inventory.py
