89
75
>>> i.add(InventoryDirectory('123', 'src', ROOT_ID))
90
InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None)
76
InventoryDirectory('123', 'src', parent_id='TREE_ROOT')
91
77
>>> i.add(InventoryFile('2323', 'hello.c', parent_id='123'))
92
InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None)
93
>>> shouldbe = {0: '', 1: 'src', 2: 'src/hello.c'}
94
>>> for ix, j in enumerate(i.iter_entries()):
95
... print (j[0] == shouldbe[ix], j[1])
78
InventoryFile('2323', 'hello.c', parent_id='123')
79
>>> for j in i.iter_entries():
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))
82
('src', InventoryDirectory('123', 'src', parent_id='TREE_ROOT'))
83
('src/hello.c', InventoryFile('2323', 'hello.c', parent_id='123'))
84
>>> i.add(InventoryFile('2323', 'bye.c', '123'))
85
Traceback (most recent call last):
87
BzrError: inventory already contains entry with id {2323}
100
88
>>> i.add(InventoryFile('2324', 'bye.c', '123'))
101
InventoryFile('2324', 'bye.c', parent_id='123', sha1=None, len=None)
89
InventoryFile('2324', 'bye.c', parent_id='123')
102
90
>>> i.add(InventoryDirectory('2325', 'wibble', '123'))
103
InventoryDirectory('2325', 'wibble', parent_id='123', revision=None)
91
InventoryDirectory('2325', 'wibble', parent_id='123')
104
92
>>> i.path2id('src/wibble')
108
96
>>> i.add(InventoryFile('2326', 'wibble.c', '2325'))
109
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
97
InventoryFile('2326', 'wibble.c', parent_id='2325')
111
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
99
InventoryFile('2326', 'wibble.c', parent_id='2325')
112
100
>>> for path, entry in i.iter_entries():
101
... print path.replace('\\\\', '/') # for win32 os.sep
114
102
... assert i.path2id(path)
121
108
src/wibble/wibble.c
122
>>> i.id2path('2326')
109
>>> i.id2path('2326').replace('\\\\', '/')
123
110
'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'
113
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
114
'text_id', 'parent_id', 'children', 'executable',
117
def _add_text_to_weave(self, new_lines, parents, weave_store, transaction):
118
weave_store.add_text(self.file_id, self.revision, new_lines, parents,
136
121
def detect_changes(self, old_entry):
137
122
"""Return a (text_modified, meta_modified) from this to old_entry.
162
147
output_to, reverse=False):
163
148
"""Perform a diff between two entries of the same kind."""
165
def find_previous_heads(self, previous_inventories,
166
versioned_file_store,
169
"""Return the revisions and entries that directly precede this.
150
def find_previous_heads(self, previous_inventories, entry_weave):
151
"""Return the revisions and entries that directly preceed this.
171
153
Returned as a map from revision to inventory entry.
173
155
This is a map containing the file revisions in all parents
174
156
for which the file exists, and its revision is not a parent of
175
157
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.
183
159
def get_ancestors(weave, entry):
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.
160
return set(map(weave.idx_to_name,
161
weave.inclusions([weave.lookup(entry.revision)])))
189
# revision: ancestor list for each head
190
163
head_ancestors = {}
191
# identify candidate head revision ids.
192
164
for inv in previous_inventories:
193
165
if self.file_id in inv:
194
166
ie = inv[self.file_id]
195
167
assert ie.file_id == self.file_id
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
168
if ie.revision in heads:
169
# fixup logic, there was a bug in revision updates.
170
# with x bit support.
205
if candidates[ie.revision].executable != ie.executable:
206
candidates[ie.revision].executable = False
172
if heads[ie.revision].executable != ie.executable:
173
heads[ie.revision].executable = False
207
174
ie.executable = False
208
175
except AttributeError:
210
# must now be the same.
211
assert candidates[ie.revision] == ie
177
assert 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
179
# may want to add it.
180
# may already be covered:
181
already_present = 0 != len(
182
[head for head in heads
183
if ie.revision in head_ancestors[head]])
185
# an ancestor of a known head.
188
ancestors = get_ancestors(entry_weave, ie)
189
# may knock something else out:
190
check_heads = list(heads.keys())
191
for head in check_heads:
192
if head in ancestors:
193
# this head is not really a head
195
head_ancestors[ie.revision] = ancestors
196
heads[ie.revision] = ie
254
199
def get_tar_item(self, root, dp, now, tree):
255
200
"""Get a tarfile item and a file stream for its content."""
256
item = tarfile.TarInfo(osutils.pathjoin(root, dp).encode('utf8'))
201
item = tarfile.TarInfo(os.path.join(root, dp))
257
202
# TODO: would be cool to actually set it to the timestamp of the
258
203
# revision it was last changed
285
230
>>> e = InventoryFile('123', 'src/hello.c', ROOT_ID)
286
231
Traceback (most recent call last):
287
InvalidEntryName: Invalid entry name: src/hello.c
232
BzrCheckError: InventoryEntry name 'src/hello.c' is invalid
289
234
assert isinstance(name, basestring), name
290
235
if '/' in name or '\\' in name:
291
raise errors.InvalidEntryName(name=name)
236
raise BzrCheckError('InventoryEntry name %r is invalid' % name)
292
238
self.executable = False
293
239
self.revision = None
294
240
self.text_sha1 = None
295
241
self.text_size = None
296
242
self.file_id = file_id
297
assert isinstance(file_id, (str, None.__class__)), \
298
'bad type %r for %r' % (type(file_id), file_id)
300
244
self.text_id = text_id
301
245
self.parent_id = parent_id
302
246
self.symlink_target = None
303
self.reference_revision = None
305
248
def kind_character(self):
306
249
"""Return a short kind indicator useful for appending to names."""
307
250
raise BzrError('unknown kind %r' % self.kind)
309
known_kinds = ('file', 'directory', 'symlink')
252
known_kinds = ('file', 'directory', 'symlink', 'root_directory')
311
254
def _put_in_tar(self, item, tree):
312
255
"""populate item for stashing in a tar, and return the content stream.
322
265
This is a template method - implement _put_on_disk in subclasses.
324
fullpath = osutils.pathjoin(dest, dp)
267
fullpath = appendpath(dest, dp)
325
268
self._put_on_disk(fullpath, tree)
326
# mutter(" export {%s} kind %s to %s", self.file_id,
327
# self.kind, fullpath)
269
mutter(" export {%s} kind %s to %s" % (self.file_id, self.kind, fullpath))
329
271
def _put_on_disk(self, fullpath, tree):
330
272
"""Put this entry onto disk at fullpath, from tree tree."""
331
273
raise BzrError("don't know how to export {%s} of kind %r" % (self.file_id, self.kind))
333
275
def sorted_children(self):
334
return sorted(self.children.items())
276
l = self.children.items()
337
281
def versionable_kind(kind):
338
return (kind in ('file', 'directory', 'symlink', 'tree-reference'))
282
return kind in ('file', 'directory', 'symlink')
340
284
def check(self, checker, rev_id, inv, tree):
341
285
"""Check this inventory entry is intact.
343
287
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.
354
if self.parent_id is not None:
290
if self.parent_id != None:
355
291
if not inv.has_id(self.parent_id):
356
292
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
357
293
% (self.parent_id, rev_id))
362
298
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
363
299
(self.kind, rev_id))
366
303
"""Clone this inventory entry."""
367
304
raise NotImplementedError
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:
306
def _get_snapshot_change(self, previous_entries):
307
if len(previous_entries) > 1:
309
elif len(previous_entries) == 0:
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
312
return 'modified/renamed/reparented'
411
314
def __repr__(self):
412
return ("%s(%r, %r, parent_id=%r, revision=%r)"
315
return ("%s(%r, %r, parent_id=%r)"
413
316
% (self.__class__.__name__,
419
321
def snapshot(self, revision, path, previous_entries,
420
work_tree, commit_builder):
322
work_tree, weave_store, transaction):
421
323
"""Make a snapshot of this entry which may or may not have changed.
423
325
This means that all its fields are populated, that it has its
424
326
text stored in the text store or weave.
426
# mutter('new parents of %s are %r', path, previous_entries)
328
mutter('new parents of %s are %r', path, previous_entries)
427
329
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
430
330
if len(previous_entries) == 1:
431
331
# cannot be unchanged unless there is only one parent file rev.
432
332
parent_ie = previous_entries.values()[0]
433
333
if self._unchanged(parent_ie):
434
# mutter("found unchanged entry")
334
mutter("found unchanged entry")
435
335
self.revision = parent_ie.revision
436
336
return "unchanged"
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)
337
return self.snapshot_revision(revision, previous_entries,
338
work_tree, weave_store, transaction)
340
def snapshot_revision(self, revision, previous_entries, work_tree,
341
weave_store, transaction):
342
"""Record this revision unconditionally."""
343
mutter('new revision for {%s}', self.file_id)
452
344
self.revision = revision
453
self._snapshot_text(previous_entries, work_tree, commit_builder)
345
change = self._get_snapshot_change(previous_entries)
346
self._snapshot_text(previous_entries, work_tree, weave_store,
455
def _snapshot_text(self, file_parents, work_tree, commit_builder):
350
def _snapshot_text(self, file_parents, work_tree, weave_store, transaction):
456
351
"""Record the 'text' of this entry, whatever form that takes.
458
353
This default implementation simply adds an empty text.
460
raise NotImplementedError(self._snapshot_text)
355
mutter('storing file {%s} in revision {%s}',
356
self.file_id, self.revision)
357
self._add_text_to_weave([], file_parents, weave_store, transaction)
462
359
def __eq__(self, other):
463
360
if not isinstance(other, InventoryEntry):
504
398
Note that this should be modified to be a noop on virtual trees
505
399
as all entries created there are prepopulated.
507
# TODO: Rather than running this manually, we should check the
508
# working sha1 and other expensive properties when they're
509
# first requested, or preload them if they're already known
510
pass # nothing to do by default
512
def _forget_tree_state(self):
516
403
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']
522
405
def _check(self, checker, rev_id, tree):
523
406
"""See InventoryEntry._check"""
525
408
def __init__(self, file_id):
526
409
self.file_id = file_id
527
410
self.children = {}
528
self.kind = 'directory'
411
self.kind = 'root_directory'
529
412
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)
536
415
def __eq__(self, other):
537
416
if not isinstance(other, RootEntry):
544
423
class InventoryDirectory(InventoryEntry):
545
424
"""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']
551
426
def _check(self, checker, rev_id, tree):
552
427
"""See InventoryEntry._check"""
553
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
428
if self.text_sha1 != None or self.text_size != None or self.text_id != None:
554
429
raise BzrCheckError('directory {%s} has text in revision {%s}'
555
430
% (self.file_id, rev_id))
583
458
"""See InventoryEntry._put_on_disk."""
584
459
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)
591
462
class InventoryFile(InventoryEntry):
592
463
"""A file in an inventory."""
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):
465
def _check(self, checker, rev_id, tree):
599
466
"""See InventoryEntry._check"""
600
t = (self.file_id, self.revision)
467
revision = self.revision
468
t = (self.file_id, revision)
601
469
if t in checker.checked_texts:
602
prev_sha = checker.checked_texts[t]
470
prev_sha = checker.checked_texts[t]
603
471
if prev_sha != self.text_sha1:
604
472
raise BzrCheckError('mismatched sha1 on {%s} in {%s}' %
605
(self.file_id, tree_revision_id))
473
(self.file_id, rev_id))
607
475
checker.repeated_text_cnt += 1
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))
477
mutter('check version {%s} of {%s}', rev_id, self.file_id)
478
file_lines = tree.get_file_lines(self.file_id)
479
checker.checked_text_cnt += 1
480
if self.text_size != sum(map(len, file_lines)):
481
raise BzrCheckError('text {%s} wrong size' % self.text_id)
482
if self.text_sha1 != sha_strings(file_lines):
483
raise BzrCheckError('text {%s} wrong sha1' % self.text_id)
629
484
checker.checked_texts[t] = self.text_sha1
648
503
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
649
504
output_to, reverse=False):
650
505
"""See InventoryEntry._diff."""
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
506
from_text = tree.get_file(self.file_id).readlines()
508
to_text = to_tree.get_file(to_entry.file_id).readlines()
512
text_diff(from_label, from_text,
513
to_label, to_text, output_to)
515
text_diff(to_label, to_text,
516
from_label, from_text, output_to)
670
518
def has_text(self):
671
519
"""See InventoryEntry.has_text."""
693
541
def _put_on_disk(self, fullpath, tree):
694
542
"""See InventoryEntry._put_on_disk."""
695
osutils.pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
543
pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
696
544
if tree.is_executable(self.file_id):
697
545
os.chmod(fullpath, 0755)
699
547
def _read_tree_state(self, path, work_tree):
700
548
"""See InventoryEntry._read_tree_state."""
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):
549
self.text_sha1 = work_tree.get_file_sha1(self.file_id)
550
self.executable = work_tree.is_executable(self.file_id)
552
def _snapshot_text(self, file_parents, work_tree, weave_store, transaction):
719
553
"""See InventoryEntry._snapshot_text."""
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)
554
mutter('storing file {%s} in revision {%s}',
555
self.file_id, self.revision)
556
# special case to avoid diffing on renames or
558
if (len(file_parents) == 1
559
and self.text_sha1 == file_parents.values()[0].text_sha1
560
and self.text_size == file_parents.values()[0].text_size):
561
previous_ie = file_parents.values()[0]
562
weave_store.add_identical_text(
563
self.file_id, previous_ie.revision,
564
self.revision, file_parents, transaction)
566
new_lines = work_tree.get_file(self.file_id).readlines()
567
self._add_text_to_weave(new_lines, file_parents, weave_store,
569
self.text_sha1 = sha_strings(new_lines)
570
self.text_size = sum(map(len, new_lines))
725
573
def _unchanged(self, previous_ie):
726
574
"""See InventoryEntry._unchanged."""
739
587
class InventoryLink(InventoryEntry):
740
588
"""A file in an inventory."""
742
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
743
'text_id', 'parent_id', 'children', 'executable',
744
'revision', 'symlink_target', 'reference_revision']
590
__slots__ = ['symlink_target']
746
592
def _check(self, checker, rev_id, tree):
747
593
"""See InventoryEntry._check"""
748
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
594
if self.text_sha1 != None or self.text_size != None or self.text_id != None:
749
595
raise BzrCheckError('symlink {%s} has text in revision {%s}'
750
596
% (self.file_id, rev_id))
751
if self.symlink_target is None:
597
if self.symlink_target == None:
752
598
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
753
599
% (self.file_id, rev_id))
822
665
compatible = False
823
666
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
858
669
class Inventory(object):
859
670
"""Inventory of versioned files in a tree.
888
699
May also look up by name:
890
701
>>> [x[0] for x in inv.iter_entries()]
892
703
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
893
704
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
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)
705
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT-12345678-12345678')
899
def __init__(self, root_id=ROOT_ID, revision_id=None):
707
def __init__(self, root_id=ROOT_ID):
900
708
"""Create or read an inventory.
902
710
If a working directory is specified, the inventory is read
906
714
The inventory is created with a default root directory, with
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):
717
# We are letting Branch.initialize() create a unique inventory
718
# root id. Rather than generating a random one here.
720
# root_id = bzrlib.branch.gen_file_id('TREE_ROOT')
721
self.root = RootEntry(root_id)
919
722
self._byid = {self.root.file_id: self.root}
922
# TODO: jam 20051218 Should copy also copy the revision_id?
923
entries = self.iter_entries()
924
other = Inventory(entries.next()[1].file_id)
726
other = Inventory(self.root.file_id)
925
727
# copy recursively so we know directories will be added before
926
728
# their children. There are more efficient ways than this...
927
for path, entry in entries():
729
for path, entry in self.iter_entries():
730
if entry == self.root:
928
732
other.add(entry.copy())
931
736
def __iter__(self):
932
737
return iter(self._byid)
934
740
def __len__(self):
935
741
"""Returns number of entries."""
936
742
return len(self._byid)
938
745
def iter_entries(self, from_dir=None):
939
746
"""Return (path, entry) pairs, in order by name."""
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))
1050
def make_entry(self, kind, name, parent_id, file_id=None):
1051
"""Simple thunk to bzrlib.inventory.make_entry."""
1052
return make_entry(kind, name, parent_id, file_id)
750
elif isinstance(from_dir, basestring):
751
from_dir = self._byid[from_dir]
753
kids = from_dir.children.items()
755
for name, ie in kids:
757
if ie.kind == 'directory':
758
for cn, cie in self.iter_entries(from_dir=ie.file_id):
759
yield os.path.join(name, cn), cie
1054
762
def entries(self):
1055
763
"""Return list of (path, ie) for all entries except the root.
1082
791
for name, child_ie in kids:
1083
child_path = osutils.pathjoin(parent_path, name)
792
child_path = os.path.join(parent_path, name)
1084
793
descend(child_ie, child_path)
1085
descend(self.root, u'')
794
descend(self.root, '')
1088
799
def __contains__(self, file_id):
1089
800
"""True if this entry contains a file with given id.
1091
802
>>> inv = Inventory()
1092
803
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1093
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
804
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT')
1094
805
>>> '123' in inv
1096
807
>>> '456' in inv
1099
file_id = osutils.safe_file_id(file_id)
1100
return (file_id in self._byid)
810
return file_id in self._byid
1102
813
def __getitem__(self, file_id):
1103
814
"""Return the entry for given file_id.
1105
816
>>> inv = Inventory()
1106
817
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
1107
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
818
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT')
1108
819
>>> inv['123123'].name
1111
file_id = osutils.safe_file_id(file_id)
1113
823
return self._byid[file_id]
1114
824
except KeyError:
1115
# really we're passing an inventory, not a tree...
1116
raise errors.NoSuchId(self, file_id)
826
raise BzrError("can't look up file_id None")
828
raise BzrError("file_id {%s} not in inventory" % file_id)
1118
831
def get_file_kind(self, file_id):
1119
file_id = osutils.safe_file_id(file_id)
1120
832
return self._byid[file_id].kind
1122
834
def get_child(self, parent_id, filename):
1123
parent_id = osutils.safe_file_id(parent_id)
1124
835
return self[parent_id].children.get(filename)
1126
def _add_child(self, entry):
1127
"""Add an entry to the inventory, without adding it to its parent"""
1128
if entry.file_id in self._byid:
1129
raise BzrError("inventory already contains entry with id {%s}" %
1131
self._byid[entry.file_id] = entry
1132
for child in getattr(entry, 'children', {}).itervalues():
1133
self._add_child(child)
1136
838
def add(self, entry):
1137
839
"""Add entry to inventory.
1142
844
Returns the new entry object.
1144
846
if entry.file_id in self._byid:
1145
raise errors.DuplicateFileId(entry.file_id,
1146
self._byid[entry.file_id])
1148
if entry.parent_id is None:
1149
assert self.root is None and len(self._byid) == 0
1153
parent = self._byid[entry.parent_id]
1155
raise BzrError("parent_id {%s} not in inventory" %
1158
if entry.name in parent.children:
1159
raise BzrError("%s is already versioned" %
1160
osutils.pathjoin(self.id2path(parent.file_id),
1161
entry.name).encode('utf-8'))
1162
parent.children[entry.name] = entry
1163
return self._add_child(entry)
1165
def add_path(self, relpath, kind, file_id=None, parent_id=None):
847
raise BzrError("inventory already contains entry with id {%s}" % entry.file_id)
849
if entry.parent_id == ROOT_ID or entry.parent_id is None:
850
entry.parent_id = self.root.file_id
853
parent = self._byid[entry.parent_id]
855
raise BzrError("parent_id {%s} not in inventory" % entry.parent_id)
857
if parent.children.has_key(entry.name):
858
raise BzrError("%s is already versioned" %
859
appendpath(self.id2path(parent.file_id), entry.name))
861
self._byid[entry.file_id] = entry
862
parent.children[entry.name] = entry
866
def add_path(self, relpath, kind, file_id=None):
1166
867
"""Add entry from a path.
1168
869
The immediate parent must already be versioned.
1170
871
Returns the new entry object."""
872
from bzrlib.branch import gen_file_id
1172
parts = osutils.splitpath(relpath)
874
parts = bzrlib.osutils.splitpath(relpath)
1174
875
if len(parts) == 0:
1176
file_id = generate_ids.gen_root_id()
1178
file_id = osutils.safe_file_id(file_id)
1179
self.root = InventoryDirectory(file_id, '', None)
1180
self._byid = {self.root.file_id: self.root}
876
raise BzrError("cannot re-add root of inventory")
879
file_id = gen_file_id(relpath)
881
parent_path = parts[:-1]
882
parent_id = self.path2id(parent_path)
883
if parent_id == None:
884
raise NotVersionedError(parent_path)
886
if kind == 'directory':
887
ie = InventoryDirectory(file_id, parts[-1], parent_id)
889
ie = InventoryFile(file_id, parts[-1], parent_id)
890
elif kind == 'symlink':
891
ie = InventoryLink(file_id, parts[-1], parent_id)
1183
parent_path = parts[:-1]
1184
parent_id = self.path2id(parent_path)
1185
if parent_id is None:
1186
raise errors.NotVersionedError(path=parent_path)
1187
ie = make_entry(kind, parts[-1], parent_id, file_id)
893
raise BzrError("unknown kind %r" % kind)
1188
894
return self.add(ie)
1190
897
def __delitem__(self, file_id):
1191
898
"""Remove entry by id.
1193
900
>>> inv = Inventory()
1194
901
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1195
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
902
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT')
1196
903
>>> '123' in inv
1198
905
>>> del inv['123']
1199
906
>>> '123' in inv
1202
file_id = osutils.safe_file_id(file_id)
1203
909
ie = self[file_id]
1205
assert ie.parent_id is None or \
1206
self[ie.parent_id].children[ie.name] == ie
911
assert self[ie.parent_id].children[ie.name] == ie
913
# TODO: Test deleting all children; maybe hoist to a separate
915
if ie.kind == 'directory':
916
for cie in ie.children.values():
917
del self[cie.file_id]
1208
920
del self._byid[file_id]
1209
if ie.parent_id is not None:
1210
del self[ie.parent_id].children[ie.name]
921
del self[ie.parent_id].children[ie.name]
1212
924
def __eq__(self, other):
1213
925
"""Compare two sets by comparing their contents.
1219
931
>>> i1.add(InventoryFile('123', 'foo', ROOT_ID))
1220
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
932
InventoryFile('123', 'foo', parent_id='TREE_ROOT')
1223
935
>>> i2.add(InventoryFile('123', 'foo', ROOT_ID))
1224
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
936
InventoryFile('123', 'foo', parent_id='TREE_ROOT')
1228
940
if not isinstance(other, Inventory):
1229
941
return NotImplemented
943
if len(self._byid) != len(other._byid):
944
# shortcut: obviously not the same
1231
947
return self._byid == other._byid
1233
950
def __ne__(self, other):
1234
951
return not self.__eq__(other)
1236
954
def __hash__(self):
1237
955
raise ValueError('not hashable')
1239
def _iter_file_id_parents(self, file_id):
1240
"""Yield the parents of file_id up to the root."""
1241
file_id = osutils.safe_file_id(file_id)
1242
while file_id is not None:
1244
ie = self._byid[file_id]
1246
raise errors.NoSuchId(tree=None, file_id=file_id)
1248
file_id = ie.parent_id
1250
958
def get_idpath(self, file_id):
1251
959
"""Return a list of file_ids for the path to an entry.
1255
963
is equal to the depth of the file in the tree, counting the
1256
964
root directory as depth 1.
1258
file_id = osutils.safe_file_id(file_id)
1260
for parent in self._iter_file_id_parents(file_id):
1261
p.insert(0, parent.file_id)
967
while file_id != None:
969
ie = self._byid[file_id]
971
raise BzrError("file_id {%s} not found in inventory" % file_id)
972
p.insert(0, ie.file_id)
973
file_id = ie.parent_id
1264
977
def id2path(self, file_id):
1265
"""Return as a string the path to file_id.
978
"""Return as a list the path to file_id.
1267
980
>>> i = Inventory()
1268
981
>>> e = i.add(InventoryDirectory('src-id', 'src', ROOT_ID))
1269
982
>>> e = i.add(InventoryFile('foo-id', 'foo.c', parent_id='src-id'))
1270
>>> print i.id2path('foo-id')
983
>>> print i.id2path('foo-id').replace(os.sep, '/')
1273
file_id = osutils.safe_file_id(file_id)
1274
986
# get all names, skipping root
1275
return '/'.join(reversed(
1276
[parent.name for parent in
1277
self._iter_file_id_parents(file_id)][:-1]))
987
p = [self._byid[fid].name for fid in self.get_idpath(file_id)[1:]]
988
return os.sep.join(p)
1279
992
def path2id(self, name):
1280
993
"""Walk down through directories to return entry of last component.
1311
1019
return parent.file_id
1313
1022
def has_filename(self, names):
1314
1023
return bool(self.path2id(names))
1316
1026
def has_id(self, file_id):
1317
file_id = osutils.safe_file_id(file_id)
1318
return (file_id in self._byid)
1027
return self._byid.has_key(file_id)
1320
def remove_recursive_id(self, file_id):
1321
"""Remove file_id, and children, from the inventory.
1323
:param file_id: A file_id to remove.
1325
file_id = osutils.safe_file_id(file_id)
1326
to_find_delete = [self._byid[file_id]]
1328
while to_find_delete:
1329
ie = to_find_delete.pop()
1330
to_delete.append(ie.file_id)
1331
if ie.kind == 'directory':
1332
to_find_delete.extend(ie.children.values())
1333
for file_id in reversed(to_delete):
1335
del self._byid[file_id]
1336
if ie.parent_id is not None:
1337
del self[ie.parent_id].children[ie.name]
1341
1030
def rename(self, file_id, new_parent_id, new_name):
1342
1031
"""Move a file within the inventory.
1344
1033
This can change either the name, or the parent, or both.
1346
This does not move the working file.
1348
file_id = osutils.safe_file_id(file_id)
1035
This does not move the working file."""
1349
1036
if not is_valid_name(new_name):
1350
1037
raise BzrError("not an acceptable filename: %r" % new_name)
1369
1056
file_ie.name = new_name
1370
1057
file_ie.parent_id = new_parent_id
1372
def is_root(self, file_id):
1373
file_id = osutils.safe_file_id(file_id)
1374
return self.root is not None and file_id == self.root.file_id
1378
'directory': InventoryDirectory,
1379
'file': InventoryFile,
1380
'symlink': InventoryLink,
1381
'tree-reference': TreeReference
1384
def make_entry(kind, name, parent_id, file_id=None):
1385
"""Create an inventory entry.
1387
:param kind: the type of inventory entry to create.
1388
:param name: the basename of the entry.
1389
:param parent_id: the parent_id of the entry.
1390
:param file_id: the file_id to use. if None, one will be created.
1393
file_id = generate_ids.gen_file_id(name)
1395
file_id = osutils.safe_file_id(file_id)
1397
#------- This has been copied to bzrlib.dirstate.DirState.add, please
1398
# keep them synchronised.
1399
# we dont import normalized_filename directly because we want to be
1400
# able to change the implementation at runtime for tests.
1401
norm_name, can_access = osutils.normalized_filename(name)
1402
if norm_name != name:
1406
# TODO: jam 20060701 This would probably be more useful
1407
# if the error was raised with the full path
1408
raise errors.InvalidNormalization(name)
1411
factory = entry_factory[kind]
1413
raise BzrError("unknown kind %r" % kind)
1414
return factory(file_id, name, parent_id)
1417
1062
_NAME_RE = None
1419
1064
def is_valid_name(name):
1420
1065
global _NAME_RE
1421
if _NAME_RE is None:
1066
if _NAME_RE == None:
1422
1067
_NAME_RE = re.compile(r'^[^/\\]+$')
1424
1069
return bool(_NAME_RE.match(name))
added
removed
bzrlib/inventory.py
