36
61
An InventoryEntry has the following fields, which are also
37
62
present in the XML inventory-entry element:
40
* *name*: (only the basename within the directory, must not
42
* *kind*: "directory" or "file"
43
* *directory_id*: (if absent/null means the branch root directory)
44
* *text_sha1*: only for files
45
* *text_size*: in bytes, only for files
46
* *text_id*: identifier for the text version, only for files
48
InventoryEntries can also exist inside a WorkingTree
49
inventory, in which case they are not yet bound to a
50
particular revision of the file. In that case the text_sha1,
51
text_size and text_id are absent.
67
(within the parent directory)
70
file_id of the parent directory, or ROOT_ID
73
the revision_id in which this variation of this file was
77
Indicates that this file should be executable on systems
81
sha-1 of the text of the file
84
size in bytes of the text of the file
86
(reading a version 4 tree created a text_id field.)
54
88
>>> i = Inventory()
57
>>> i.add(InventoryEntry('123', 'src', 'directory', ROOT_ID))
58
>>> i.add(InventoryEntry('2323', 'hello.c', 'file', parent_id='123'))
59
>>> for j in i.iter_entries():
62
('src', InventoryEntry('123', 'src', kind='directory', parent_id='TREE_ROOT'))
63
('src/hello.c', InventoryEntry('2323', 'hello.c', kind='file', parent_id='123'))
64
>>> i.add(InventoryEntry('2323', 'bye.c', 'file', '123'))
65
Traceback (most recent call last):
91
>>> i.add(InventoryDirectory('123', 'src', ROOT_ID))
92
InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None)
93
>>> i.add(InventoryFile('2323', 'hello.c', parent_id='123'))
94
InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None, revision=None)
95
>>> shouldbe = {0: '', 1: 'src', 2: 'src/hello.c'}
96
>>> for ix, j in enumerate(i.iter_entries()):
97
... print (j[0] == shouldbe[ix], j[1])
67
BzrError: inventory already contains entry with id {2323}
68
>>> i.add(InventoryEntry('2324', 'bye.c', 'file', '123'))
69
>>> i.add(InventoryEntry('2325', 'wibble', 'directory', '123'))
99
(True, InventoryDirectory('TREE_ROOT', u'', parent_id=None, revision=None))
100
(True, InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None))
101
(True, InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None, revision=None))
102
>>> i.add(InventoryFile('2324', 'bye.c', '123'))
103
InventoryFile('2324', 'bye.c', parent_id='123', sha1=None, len=None, revision=None)
104
>>> i.add(InventoryDirectory('2325', 'wibble', '123'))
105
InventoryDirectory('2325', 'wibble', parent_id='123', revision=None)
70
106
>>> i.path2id('src/wibble')
74
>>> i.add(InventoryEntry('2326', 'wibble.c', 'file', '2325'))
110
>>> i.add(InventoryFile('2326', 'wibble.c', '2325'))
111
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None, revision=None)
76
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
77
>>> for j in i.iter_entries():
79
... assert i.path2id(j[0])
113
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None, revision=None)
114
>>> for path, entry in i.iter_entries():
85
122
src/wibble/wibble.c
86
123
>>> i.id2path('2326')
87
124
'src/wibble/wibble.c'
89
TODO: Maybe also keep the full path of the entry, and the children?
90
But those depend on its position within a particular inventory, and
91
it would be nice not to need to hold the backpointer here.
94
# TODO: split InventoryEntry into subclasses for files,
95
# directories, etc etc.
97
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
98
'text_id', 'parent_id', 'children', ]
100
def __init__(self, file_id, name, kind, parent_id, text_id=None):
127
# Constants returned by describe_change()
129
# TODO: These should probably move to some kind of FileChangeDescription
130
# class; that's like what's inside a TreeDelta but we want to be able to
131
# generate them just for one file at a time.
133
MODIFIED_AND_RENAMED = 'modified and renamed'
137
def detect_changes(self, old_entry):
138
"""Return a (text_modified, meta_modified) from this to old_entry.
140
_read_tree_state must have been called on self and old_entry prior to
141
calling detect_changes.
145
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
146
output_to, reverse=False):
147
"""Perform a diff between two entries of the same kind."""
149
def parent_candidates(self, previous_inventories):
150
"""Find possible per-file graph parents.
152
This is currently defined by:
153
- Select the last changed revision in the parent inventory.
154
- Do deal with a short lived bug in bzr 0.8's development two entries
155
that have the same last changed but different 'x' bit settings are
158
# revision:ie mapping for each ie found in previous_inventories.
160
# identify candidate head revision ids.
161
for inv in previous_inventories:
162
if self.file_id in inv:
163
ie = inv[self.file_id]
164
if ie.revision in candidates:
165
# same revision value in two different inventories:
166
# correct possible inconsistencies:
167
# * there was a bug in revision updates with 'x' bit
170
if candidates[ie.revision].executable != ie.executable:
171
candidates[ie.revision].executable = False
172
ie.executable = False
173
except AttributeError:
176
# add this revision as a candidate.
177
candidates[ie.revision] = ie
180
@deprecated_method(deprecated_in((1, 6, 0)))
181
def get_tar_item(self, root, dp, now, tree):
182
"""Get a tarfile item and a file stream for its content."""
183
item = tarfile.TarInfo(osutils.pathjoin(root, dp).encode('utf8'))
184
# TODO: would be cool to actually set it to the timestamp of the
185
# revision it was last changed
187
fileobj = self._put_in_tar(item, tree)
191
"""Return true if the object this entry represents has textual data.
193
Note that textual data includes binary content.
195
Also note that all entries get weave files created for them.
196
This attribute is primarily used when upgrading from old trees that
197
did not have the weave index for all inventory entries.
201
def __init__(self, file_id, name, parent_id, text_id=None):
101
202
"""Create an InventoryEntry
103
204
The filename must be a single component, relative to the
104
205
parent directory; it cannot be a whole path or relative name.
106
>>> e = InventoryEntry('123', 'hello.c', 'file', ROOT_ID)
207
>>> e = InventoryFile('123', 'hello.c', ROOT_ID)
111
>>> e = InventoryEntry('123', 'src/hello.c', 'file', ROOT_ID)
212
>>> e = InventoryFile('123', 'src/hello.c', ROOT_ID)
112
213
Traceback (most recent call last):
113
BzrCheckError: InventoryEntry name 'src/hello.c' is invalid
214
InvalidEntryName: Invalid entry name: src/hello.c
115
216
if '/' in name or '\\' in name:
116
raise BzrCheckError('InventoryEntry name %r is invalid' % name)
217
raise errors.InvalidEntryName(name=name)
218
self.executable = False
118
220
self.text_sha1 = None
119
221
self.text_size = None
121
222
self.file_id = file_id
124
224
self.text_id = text_id
125
225
self.parent_id = parent_id
126
if kind == 'directory':
131
raise BzrError("unhandled entry kind %r" % kind)
226
self.symlink_target = None
227
self.reference_revision = None
229
def kind_character(self):
230
"""Return a short kind indicator useful for appending to names."""
231
raise BzrError('unknown kind %r' % self.kind)
233
known_kinds = ('file', 'directory', 'symlink')
235
def _put_in_tar(self, item, tree):
236
"""populate item for stashing in a tar, and return the content stream.
238
If no content is available, return None.
240
raise BzrError("don't know how to export {%s} of kind %r" %
241
(self.file_id, self.kind))
243
@deprecated_method(deprecated_in((1, 6, 0)))
244
def put_on_disk(self, dest, dp, tree):
245
"""Create a representation of self on disk in the prefix dest.
247
This is a template method - implement _put_on_disk in subclasses.
249
fullpath = osutils.pathjoin(dest, dp)
250
self._put_on_disk(fullpath, tree)
251
# mutter(" export {%s} kind %s to %s", self.file_id,
252
# self.kind, fullpath)
254
def _put_on_disk(self, fullpath, tree):
255
"""Put this entry onto disk at fullpath, from tree tree."""
256
raise BzrError("don't know how to export {%s} of kind %r" % (self.file_id, self.kind))
135
258
def sorted_children(self):
136
l = self.children.items()
259
return sorted(self.children.items())
262
def versionable_kind(kind):
263
return (kind in ('file', 'directory', 'symlink', 'tree-reference'))
265
def check(self, checker, rev_id, inv, tree):
266
"""Check this inventory entry is intact.
268
This is a template method, override _check for kind specific
271
:param checker: Check object providing context for the checks;
272
can be used to find out what parts of the repository have already
274
:param rev_id: Revision id from which this InventoryEntry was loaded.
275
Not necessarily the last-changed revision for this file.
276
:param inv: Inventory from which the entry was loaded.
277
:param tree: RevisionTree for this entry.
279
if self.parent_id is not None:
280
if not inv.has_id(self.parent_id):
281
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
282
% (self.parent_id, rev_id))
283
self._check(checker, rev_id, tree)
285
def _check(self, checker, rev_id, tree):
286
"""Check this inventory entry for kind specific errors."""
287
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
142
other = InventoryEntry(self.file_id, self.name, self.kind,
143
self.parent_id, text_id=self.text_id)
144
other.text_sha1 = self.text_sha1
145
other.text_size = self.text_size
146
# note that children are *not* copied; they're pulled across when
291
"""Clone this inventory entry."""
292
raise NotImplementedError
295
def describe_change(old_entry, new_entry):
296
"""Describe the change between old_entry and this.
298
This smells of being an InterInventoryEntry situation, but as its
299
the first one, we're making it a static method for now.
301
An entry with a different parent, or different name is considered
302
to be renamed. Reparenting is an internal detail.
303
Note that renaming the parent does not trigger a rename for the
306
# TODO: Perhaps return an object rather than just a string
307
if old_entry is new_entry:
308
# also the case of both being None
310
elif old_entry is None:
312
elif new_entry is None:
314
if old_entry.kind != new_entry.kind:
316
text_modified, meta_modified = new_entry.detect_changes(old_entry)
317
if text_modified or meta_modified:
321
# TODO 20060511 (mbp, rbc) factor out 'detect_rename' here.
322
if old_entry.parent_id != new_entry.parent_id:
324
elif old_entry.name != new_entry.name:
328
if renamed and not modified:
329
return InventoryEntry.RENAMED
330
if modified and not renamed:
332
if modified and renamed:
333
return InventoryEntry.MODIFIED_AND_RENAMED
151
336
def __repr__(self):
152
return ("%s(%r, %r, kind=%r, parent_id=%r)"
337
return ("%s(%r, %r, parent_id=%r, revision=%r)"
153
338
% (self.__class__.__name__,
160
def to_element(self):
161
"""Convert to XML element"""
162
from bzrlib.xml import Element
166
e.set('name', self.name)
167
e.set('file_id', self.file_id)
168
e.set('kind', self.kind)
170
if self.text_size != None:
171
e.set('text_size', '%d' % self.text_size)
173
for f in ['text_id', 'text_sha1']:
178
# to be conservative, we don't externalize the root pointers
179
# for now, leaving them as null in the xml form. in a future
180
# version it will be implied by nested elements.
181
if self.parent_id != ROOT_ID:
182
assert isinstance(self.parent_id, basestring)
183
e.set('parent_id', self.parent_id)
190
def from_element(cls, elt):
191
assert elt.tag == 'entry'
193
## original format inventories don't have a parent_id for
194
## nodes in the root directory, but it's cleaner to use one
196
parent_id = elt.get('parent_id')
197
if parent_id == None:
200
self = cls(elt.get('file_id'), elt.get('name'), elt.get('kind'), parent_id)
201
self.text_id = elt.get('text_id')
202
self.text_sha1 = elt.get('text_sha1')
204
## mutter("read inventoryentry: %r" % (elt.attrib))
206
v = elt.get('text_size')
207
self.text_size = v and int(v)
212
from_element = classmethod(from_element)
214
344
def __eq__(self, other):
346
# For the case when objects are cached
215
348
if not isinstance(other, InventoryEntry):
216
349
return NotImplemented
218
return (self.file_id == other.file_id) \
219
and (self.name == other.name) \
220
and (self.text_sha1 == other.text_sha1) \
221
and (self.text_size == other.text_size) \
222
and (self.text_id == other.text_id) \
223
and (self.parent_id == other.parent_id) \
224
and (self.kind == other.kind)
351
return ((self.file_id == other.file_id)
352
and (self.name == other.name)
353
and (other.symlink_target == self.symlink_target)
354
and (self.text_sha1 == other.text_sha1)
355
and (self.text_size == other.text_size)
356
and (self.text_id == other.text_id)
357
and (self.parent_id == other.parent_id)
358
and (self.kind == other.kind)
359
and (self.revision == other.revision)
360
and (self.executable == other.executable)
361
and (self.reference_revision == other.reference_revision)
227
364
def __ne__(self, other):
228
365
return not (self == other)
230
367
def __hash__(self):
231
368
raise ValueError('not hashable')
370
def _unchanged(self, previous_ie):
371
"""Has this entry changed relative to previous_ie.
373
This method should be overridden in child classes.
376
# different inv parent
377
if previous_ie.parent_id != self.parent_id:
380
elif previous_ie.name != self.name:
382
elif previous_ie.kind != self.kind:
386
def _read_tree_state(self, path, work_tree):
387
"""Populate fields in the inventory entry from the given tree.
389
Note that this should be modified to be a noop on virtual trees
390
as all entries created there are prepopulated.
392
# TODO: Rather than running this manually, we should check the
393
# working sha1 and other expensive properties when they're
394
# first requested, or preload them if they're already known
395
pass # nothing to do by default
397
def _forget_tree_state(self):
235
401
class RootEntry(InventoryEntry):
403
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
404
'text_id', 'parent_id', 'children', 'executable',
405
'revision', 'symlink_target', 'reference_revision']
407
def _check(self, checker, rev_id, tree):
408
"""See InventoryEntry._check"""
236
410
def __init__(self, file_id):
237
411
self.file_id = file_id
238
412
self.children = {}
239
self.kind = 'root_directory'
413
self.kind = 'directory'
240
414
self.parent_id = None
417
symbol_versioning.warn('RootEntry is deprecated as of bzr 0.10.'
418
' Please use InventoryDirectory instead.',
419
DeprecationWarning, stacklevel=2)
243
421
def __eq__(self, other):
244
422
if not isinstance(other, RootEntry):
245
423
return NotImplemented
247
425
return (self.file_id == other.file_id) \
248
426
and (self.children == other.children)
252
class Inventory(object):
429
class InventoryDirectory(InventoryEntry):
430
"""A directory in an inventory."""
432
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
433
'text_id', 'parent_id', 'children', 'executable',
434
'revision', 'symlink_target', 'reference_revision']
436
def _check(self, checker, rev_id, tree):
437
"""See InventoryEntry._check"""
438
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
439
raise BzrCheckError('directory {%s} has text in revision {%s}'
440
% (self.file_id, rev_id))
443
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
444
other.revision = self.revision
445
# note that children are *not* copied; they're pulled across when
449
def __init__(self, file_id, name, parent_id):
450
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
452
self.kind = 'directory'
454
def kind_character(self):
455
"""See InventoryEntry.kind_character."""
458
def _put_in_tar(self, item, tree):
459
"""See InventoryEntry._put_in_tar."""
460
item.type = tarfile.DIRTYPE
467
def _put_on_disk(self, fullpath, tree):
468
"""See InventoryEntry._put_on_disk."""
472
class InventoryFile(InventoryEntry):
473
"""A file in an inventory."""
475
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
476
'text_id', 'parent_id', 'children', 'executable',
477
'revision', 'symlink_target', 'reference_revision']
479
def _check(self, checker, tree_revision_id, tree):
480
"""See InventoryEntry._check"""
481
key = (self.file_id, self.revision)
482
if key in checker.checked_texts:
483
prev_sha = checker.checked_texts[key]
484
if prev_sha != self.text_sha1:
486
'mismatched sha1 on {%s} in {%s} (%s != %s) %r' %
487
(self.file_id, tree_revision_id, prev_sha, self.text_sha1,
490
checker.repeated_text_cnt += 1
493
checker.checked_text_cnt += 1
494
# We can't check the length, because Weave doesn't store that
495
# information, and the whole point of looking at the weave's
496
# sha1sum is that we don't have to extract the text.
497
if (self.text_sha1 != tree._repository.texts.get_sha1s([key])[key]):
498
raise BzrCheckError('text {%s} version {%s} wrong sha1' % key)
499
checker.checked_texts[key] = self.text_sha1
502
other = InventoryFile(self.file_id, self.name, self.parent_id)
503
other.executable = self.executable
504
other.text_id = self.text_id
505
other.text_sha1 = self.text_sha1
506
other.text_size = self.text_size
507
other.revision = self.revision
510
def detect_changes(self, old_entry):
511
"""See InventoryEntry.detect_changes."""
512
text_modified = (self.text_sha1 != old_entry.text_sha1)
513
meta_modified = (self.executable != old_entry.executable)
514
return text_modified, meta_modified
516
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
517
output_to, reverse=False):
518
"""See InventoryEntry._diff."""
519
from bzrlib.diff import DiffText
520
from_file_id = self.file_id
522
to_file_id = to_entry.file_id
526
to_file_id, from_file_id = from_file_id, to_file_id
527
tree, to_tree = to_tree, tree
528
from_label, to_label = to_label, from_label
529
differ = DiffText(tree, to_tree, output_to, 'utf-8', '', '',
531
return differ.diff_text(from_file_id, to_file_id, from_label, to_label)
534
"""See InventoryEntry.has_text."""
537
def __init__(self, file_id, name, parent_id):
538
super(InventoryFile, self).__init__(file_id, name, parent_id)
541
def kind_character(self):
542
"""See InventoryEntry.kind_character."""
545
def _put_in_tar(self, item, tree):
546
"""See InventoryEntry._put_in_tar."""
547
item.type = tarfile.REGTYPE
548
fileobj = tree.get_file(self.file_id)
549
item.size = self.text_size
550
if tree.is_executable(self.file_id):
556
def _put_on_disk(self, fullpath, tree):
557
"""See InventoryEntry._put_on_disk."""
558
osutils.pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
559
if tree.is_executable(self.file_id):
560
os.chmod(fullpath, 0755)
562
def _read_tree_state(self, path, work_tree):
563
"""See InventoryEntry._read_tree_state."""
564
self.text_sha1 = work_tree.get_file_sha1(self.file_id, path=path)
565
# FIXME: 20050930 probe for the text size when getting sha1
566
# in _read_tree_state
567
self.executable = work_tree.is_executable(self.file_id, path=path)
570
return ("%s(%r, %r, parent_id=%r, sha1=%r, len=%s, revision=%s)"
571
% (self.__class__.__name__,
579
def _forget_tree_state(self):
580
self.text_sha1 = None
582
def _unchanged(self, previous_ie):
583
"""See InventoryEntry._unchanged."""
584
compatible = super(InventoryFile, self)._unchanged(previous_ie)
585
if self.text_sha1 != previous_ie.text_sha1:
588
# FIXME: 20050930 probe for the text size when getting sha1
589
# in _read_tree_state
590
self.text_size = previous_ie.text_size
591
if self.executable != previous_ie.executable:
596
class InventoryLink(InventoryEntry):
597
"""A file in an inventory."""
599
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
600
'text_id', 'parent_id', 'children', 'executable',
601
'revision', 'symlink_target', 'reference_revision']
603
def _check(self, checker, rev_id, tree):
604
"""See InventoryEntry._check"""
605
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
606
raise BzrCheckError('symlink {%s} has text in revision {%s}'
607
% (self.file_id, rev_id))
608
if self.symlink_target is None:
609
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
610
% (self.file_id, rev_id))
613
other = InventoryLink(self.file_id, self.name, self.parent_id)
614
other.symlink_target = self.symlink_target
615
other.revision = self.revision
618
def detect_changes(self, old_entry):
619
"""See InventoryEntry.detect_changes."""
620
# FIXME: which _modified field should we use ? RBC 20051003
621
text_modified = (self.symlink_target != old_entry.symlink_target)
623
mutter(" symlink target changed")
624
meta_modified = False
625
return text_modified, meta_modified
627
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
628
output_to, reverse=False):
629
"""See InventoryEntry._diff."""
630
from bzrlib.diff import DiffSymlink
631
old_target = self.symlink_target
632
if to_entry is not None:
633
new_target = to_entry.symlink_target
642
new_target, old_target = old_target, new_target
643
differ = DiffSymlink(old_tree, new_tree, output_to)
644
return differ.diff_symlink(old_target, new_target)
646
def __init__(self, file_id, name, parent_id):
647
super(InventoryLink, self).__init__(file_id, name, parent_id)
648
self.kind = 'symlink'
650
def kind_character(self):
651
"""See InventoryEntry.kind_character."""
654
def _put_in_tar(self, item, tree):
655
"""See InventoryEntry._put_in_tar."""
656
item.type = tarfile.SYMTYPE
660
item.linkname = self.symlink_target
663
def _put_on_disk(self, fullpath, tree):
664
"""See InventoryEntry._put_on_disk."""
666
os.symlink(self.symlink_target, fullpath)
668
raise BzrError("Failed to create symlink %r -> %r, error: %s" % (fullpath, self.symlink_target, e))
670
def _read_tree_state(self, path, work_tree):
671
"""See InventoryEntry._read_tree_state."""
672
self.symlink_target = work_tree.get_symlink_target(self.file_id)
674
def _forget_tree_state(self):
675
self.symlink_target = None
677
def _unchanged(self, previous_ie):
678
"""See InventoryEntry._unchanged."""
679
compatible = super(InventoryLink, self)._unchanged(previous_ie)
680
if self.symlink_target != previous_ie.symlink_target:
685
class TreeReference(InventoryEntry):
687
kind = 'tree-reference'
689
def __init__(self, file_id, name, parent_id, revision=None,
690
reference_revision=None):
691
InventoryEntry.__init__(self, file_id, name, parent_id)
692
self.revision = revision
693
self.reference_revision = reference_revision
696
return TreeReference(self.file_id, self.name, self.parent_id,
697
self.revision, self.reference_revision)
699
def _read_tree_state(self, path, work_tree):
700
"""Populate fields in the inventory entry from the given tree.
702
self.reference_revision = work_tree.get_reference_revision(
705
def _forget_tree_state(self):
706
self.reference_revision = None
708
def _unchanged(self, previous_ie):
709
"""See InventoryEntry._unchanged."""
710
compatible = super(TreeReference, self)._unchanged(previous_ie)
711
if self.reference_revision != previous_ie.reference_revision:
716
class CommonInventory(object):
717
"""Basic inventory logic, defined in terms of primitives like has_id."""
719
def __contains__(self, file_id):
720
"""True if this entry contains a file with given id.
722
>>> inv = Inventory()
723
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
724
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
730
Note that this method along with __iter__ are not encouraged for use as
731
they are less clear than specific query methods - they may be rmeoved
734
return self.has_id(file_id)
736
def id2path(self, file_id):
737
"""Return as a string the path to file_id.
740
>>> e = i.add(InventoryDirectory('src-id', 'src', ROOT_ID))
741
>>> e = i.add(InventoryFile('foo-id', 'foo.c', parent_id='src-id'))
742
>>> print i.id2path('foo-id')
745
# get all names, skipping root
746
return '/'.join(reversed(
747
[parent.name for parent in
748
self._iter_file_id_parents(file_id)][:-1]))
750
def iter_entries(self, from_dir=None, recursive=True):
751
"""Return (path, entry) pairs, in order by name.
753
:param from_dir: if None, start from the root,
754
otherwise start from this directory (either file-id or entry)
755
:param recursive: recurse into directories or not
758
if self.root is None:
762
elif isinstance(from_dir, basestring):
763
from_dir = self[from_dir]
765
# unrolling the recursive called changed the time from
766
# 440ms/663ms (inline/total) to 116ms/116ms
767
children = from_dir.children.items()
770
for name, ie in children:
773
children = collections.deque(children)
774
stack = [(u'', children)]
776
from_dir_relpath, children = stack[-1]
779
name, ie = children.popleft()
781
# we know that from_dir_relpath never ends in a slash
782
# and 'f' doesn't begin with one, we can do a string op, rather
783
# than the checks of pathjoin(), though this means that all paths
785
path = from_dir_relpath + '/' + name
789
if ie.kind != 'directory':
792
# But do this child first
793
new_children = ie.children.items()
795
new_children = collections.deque(new_children)
796
stack.append((path, new_children))
797
# Break out of inner loop, so that we start outer loop with child
800
# if we finished all children, pop it off the stack
803
def iter_entries_by_dir(self, from_dir=None, specific_file_ids=None,
804
yield_parents=False):
805
"""Iterate over the entries in a directory first order.
807
This returns all entries for a directory before returning
808
the entries for children of a directory. This is not
809
lexicographically sorted order, and is a hybrid between
810
depth-first and breadth-first.
812
:param yield_parents: If True, yield the parents from the root leading
813
down to specific_file_ids that have been requested. This has no
814
impact if specific_file_ids is None.
815
:return: This yields (path, entry) pairs
817
if specific_file_ids and not isinstance(specific_file_ids, set):
818
specific_file_ids = set(specific_file_ids)
819
# TODO? Perhaps this should return the from_dir so that the root is
820
# yielded? or maybe an option?
822
if self.root is None:
824
# Optimize a common case
825
if (not yield_parents and specific_file_ids is not None and
826
len(specific_file_ids) == 1):
827
file_id = list(specific_file_ids)[0]
829
yield self.id2path(file_id), self[file_id]
832
if (specific_file_ids is None or yield_parents or
833
self.root.file_id in specific_file_ids):
835
elif isinstance(from_dir, basestring):
836
from_dir = self[from_dir]
838
if specific_file_ids is not None:
839
# TODO: jam 20070302 This could really be done as a loop rather
840
# than a bunch of recursive calls.
843
def add_ancestors(file_id):
844
if file_id not in byid:
846
parent_id = byid[file_id].parent_id
847
if parent_id is None:
849
if parent_id not in parents:
850
parents.add(parent_id)
851
add_ancestors(parent_id)
852
for file_id in specific_file_ids:
853
add_ancestors(file_id)
857
stack = [(u'', from_dir)]
859
cur_relpath, cur_dir = stack.pop()
862
for child_name, child_ie in sorted(cur_dir.children.iteritems()):
864
child_relpath = cur_relpath + child_name
866
if (specific_file_ids is None or
867
child_ie.file_id in specific_file_ids or
868
(yield_parents and child_ie.file_id in parents)):
869
yield child_relpath, child_ie
871
if child_ie.kind == 'directory':
872
if parents is None or child_ie.file_id in parents:
873
child_dirs.append((child_relpath+'/', child_ie))
874
stack.extend(reversed(child_dirs))
876
def _make_delta(self, old):
877
"""Make an inventory delta from two inventories."""
880
adds = new_ids - old_ids
881
deletes = old_ids - new_ids
882
common = old_ids.intersection(new_ids)
884
for file_id in deletes:
885
delta.append((old.id2path(file_id), None, file_id, None))
887
delta.append((None, self.id2path(file_id), file_id, self[file_id]))
888
for file_id in common:
889
if old[file_id] != self[file_id]:
890
delta.append((old.id2path(file_id), self.id2path(file_id),
891
file_id, self[file_id]))
894
def _get_mutable_inventory(self):
895
"""Returns a mutable copy of the object.
897
Some inventories are immutable, yet working trees, for example, needs
898
to mutate exisiting inventories instead of creating a new one.
900
raise NotImplementedError(self._get_mutable_inventory)
902
def make_entry(self, kind, name, parent_id, file_id=None):
903
"""Simple thunk to bzrlib.inventory.make_entry."""
904
return make_entry(kind, name, parent_id, file_id)
907
"""Return list of (path, ie) for all entries except the root.
909
This may be faster than iter_entries.
912
def descend(dir_ie, dir_path):
913
kids = dir_ie.children.items()
915
for name, ie in kids:
916
child_path = osutils.pathjoin(dir_path, name)
917
accum.append((child_path, ie))
918
if ie.kind == 'directory':
919
descend(ie, child_path)
921
descend(self.root, u'')
924
def directories(self):
925
"""Return (path, entry) pairs for all directories, including the root.
928
def descend(parent_ie, parent_path):
929
accum.append((parent_path, parent_ie))
931
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
934
for name, child_ie in kids:
935
child_path = osutils.pathjoin(parent_path, name)
936
descend(child_ie, child_path)
937
descend(self.root, u'')
940
def path2id(self, name):
941
"""Walk down through directories to return entry of last component.
943
names may be either a list of path components, or a single
944
string, in which case it is automatically split.
946
This returns the entry of the last component in the path,
947
which may be either a file or a directory.
949
Returns None IFF the path is not found.
951
if isinstance(name, basestring):
952
name = osutils.splitpath(name)
954
# mutter("lookup path %r" % name)
958
except errors.NoSuchId:
959
# root doesn't exist yet so nothing else can
965
children = getattr(parent, 'children', None)
974
return parent.file_id
976
def filter(self, specific_fileids):
977
"""Get an inventory view filtered against a set of file-ids.
979
Children of directories and parents are included.
981
The result may or may not reference the underlying inventory
982
so it should be treated as immutable.
984
interesting_parents = set()
985
for fileid in specific_fileids:
987
interesting_parents.update(self.get_idpath(fileid))
988
except errors.NoSuchId:
989
# This fileid is not in the inventory - that's ok
991
entries = self.iter_entries()
992
if self.root is None:
993
return Inventory(root_id=None)
994
other = Inventory(entries.next()[1].file_id)
995
other.root.revision = self.root.revision
996
other.revision_id = self.revision_id
997
directories_to_expand = set()
998
for path, entry in entries:
999
file_id = entry.file_id
1000
if (file_id in specific_fileids
1001
or entry.parent_id in directories_to_expand):
1002
if entry.kind == 'directory':
1003
directories_to_expand.add(file_id)
1004
elif file_id not in interesting_parents:
1006
other.add(entry.copy())
1009
def get_idpath(self, file_id):
1010
"""Return a list of file_ids for the path to an entry.
1012
The list contains one element for each directory followed by
1013
the id of the file itself. So the length of the returned list
1014
is equal to the depth of the file in the tree, counting the
1015
root directory as depth 1.
1018
for parent in self._iter_file_id_parents(file_id):
1019
p.insert(0, parent.file_id)
1023
class Inventory(CommonInventory):
253
1024
"""Inventory of versioned files in a tree.
255
1026
This describes which file_id is present at each point in the tree,
295
1071
The inventory is created with a default root directory, with
298
# We are letting Branch(init=True) create a unique inventory
299
# root id. Rather than generating a random one here.
301
# root_id = bzrlib.branch.gen_file_id('TREE_ROOT')
302
self.root = RootEntry(root_id)
1074
if root_id is not None:
1075
self._set_root(InventoryDirectory(root_id, u'', None))
1079
self.revision_id = revision_id
1082
# More than one page of ouput is not useful anymore to debug
1085
contents = repr(self._byid)
1086
if len(contents) > max_len:
1087
contents = contents[:(max_len-len(closing))] + closing
1088
return "<Inventory object at %x, contents=%r>" % (id(self), contents)
1090
def apply_delta(self, delta):
1091
"""Apply a delta to this inventory.
1093
:param delta: A list of changes to apply. After all the changes are
1094
applied the final inventory must be internally consistent, but it
1095
is ok to supply changes which, if only half-applied would have an
1096
invalid result - such as supplying two changes which rename two
1097
files, 'A' and 'B' with each other : [('A', 'B', 'A-id', a_entry),
1098
('B', 'A', 'B-id', b_entry)].
1100
Each change is a tuple, of the form (old_path, new_path, file_id,
1103
When new_path is None, the change indicates the removal of an entry
1104
from the inventory and new_entry will be ignored (using None is
1105
appropriate). If new_path is not None, then new_entry must be an
1106
InventoryEntry instance, which will be incorporated into the
1107
inventory (and replace any existing entry with the same file id).
1109
When old_path is None, the change indicates the addition of
1110
a new entry to the inventory.
1112
When neither new_path nor old_path are None, the change is a
1113
modification to an entry, such as a rename, reparent, kind change
1116
The children attribute of new_entry is ignored. This is because
1117
this method preserves children automatically across alterations to
1118
the parent of the children, and cases where the parent id of a
1119
child is changing require the child to be passed in as a separate
1120
change regardless. E.g. in the recursive deletion of a directory -
1121
the directory's children must be included in the delta, or the
1122
final inventory will be invalid.
1124
Note that a file_id must only appear once within a given delta.
1125
An AssertionError is raised otherwise.
1127
# Check that the delta is legal. It would be nice if this could be
1128
# done within the loops below but it's safer to validate the delta
1129
# before starting to mutate the inventory.
1130
unique_file_ids = set([f for _, _, f, _ in delta])
1131
if len(unique_file_ids) != len(delta):
1132
raise AssertionError("a file-id appears multiple times in %r"
1137
# Remove all affected items which were in the original inventory,
1138
# starting with the longest paths, thus ensuring parents are examined
1139
# after their children, which means that everything we examine has no
1140
# modified children remaining by the time we examine it.
1141
for old_path, file_id in sorted(((op, f) for op, np, f, e in delta
1142
if op is not None), reverse=True):
1143
if file_id not in self:
1146
# Preserve unaltered children of file_id for later reinsertion.
1147
file_id_children = getattr(self[file_id], 'children', {})
1148
if len(file_id_children):
1149
children[file_id] = file_id_children
1150
# Remove file_id and the unaltered children. If file_id is not
1151
# being deleted it will be reinserted back later.
1152
self.remove_recursive_id(file_id)
1153
# Insert all affected which should be in the new inventory, reattaching
1154
# their children if they had any. This is done from shortest path to
1155
# longest, ensuring that items which were modified and whose parents in
1156
# the resulting inventory were also modified, are inserted after their
1158
for new_path, new_entry in sorted((np, e) for op, np, f, e in
1159
delta if np is not None):
1160
if new_entry.kind == 'directory':
1161
# Pop the child which to allow detection of children whose
1162
# parents were deleted and which were not reattached to a new
1164
replacement = InventoryDirectory(new_entry.file_id,
1165
new_entry.name, new_entry.parent_id)
1166
replacement.revision = new_entry.revision
1167
replacement.children = children.pop(replacement.file_id, {})
1168
new_entry = replacement
1171
# Get the parent id that was deleted
1172
parent_id, children = children.popitem()
1173
raise errors.InconsistentDelta("<deleted>", parent_id,
1174
"The file id was deleted but its children were not deleted.")
1176
def _set_root(self, ie):
303
1178
self._byid = {self.root.file_id: self.root}
1181
# TODO: jam 20051218 Should copy also copy the revision_id?
1182
entries = self.iter_entries()
1183
if self.root is None:
1184
return Inventory(root_id=None)
1185
other = Inventory(entries.next()[1].file_id)
1186
other.root.revision = self.root.revision
1187
# copy recursively so we know directories will be added before
1188
# their children. There are more efficient ways than this...
1189
for path, entry in entries:
1190
other.add(entry.copy())
1193
def _get_mutable_inventory(self):
1194
"""See CommonInventory._get_mutable_inventory."""
1195
return deepcopy(self)
306
1197
def __iter__(self):
1198
"""Iterate over all file-ids."""
307
1199
return iter(self._byid)
1201
def iter_just_entries(self):
1202
"""Iterate over all entries.
1204
Unlike iter_entries(), just the entries are returned (not (path, ie))
1205
and the order of entries is undefined.
1207
XXX: We may not want to merge this into bzr.dev.
1209
if self.root is None:
1211
for _, ie in self._byid.iteritems():
310
1214
def __len__(self):
311
1215
"""Returns number of entries."""
312
1216
return len(self._byid)
315
def iter_entries(self, from_dir=None):
316
"""Return (path, entry) pairs, in order by name."""
320
elif isinstance(from_dir, basestring):
321
from_dir = self._byid[from_dir]
323
kids = from_dir.children.items()
325
for name, ie in kids:
327
if ie.kind == 'directory':
328
for cn, cie in self.iter_entries(from_dir=ie.file_id):
329
yield os.path.join(name, cn), cie
333
"""Return list of (path, ie) for all entries except the root.
335
This may be faster than iter_entries.
338
def descend(dir_ie, dir_path):
339
kids = dir_ie.children.items()
341
for name, ie in kids:
342
child_path = os.path.join(dir_path, name)
343
accum.append((child_path, ie))
344
if ie.kind == 'directory':
345
descend(ie, child_path)
347
descend(self.root, '')
351
def directories(self):
352
"""Return (path, entry) pairs for all directories, including the root.
355
def descend(parent_ie, parent_path):
356
accum.append((parent_path, parent_ie))
358
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
361
for name, child_ie in kids:
362
child_path = os.path.join(parent_path, name)
363
descend(child_ie, child_path)
364
descend(self.root, '')
369
def __contains__(self, file_id):
370
"""True if this entry contains a file with given id.
372
>>> inv = Inventory()
373
>>> inv.add(InventoryEntry('123', 'foo.c', 'file', ROOT_ID))
379
return file_id in self._byid
382
1218
def __getitem__(self, file_id):
383
1219
"""Return the entry for given file_id.
385
1221
>>> inv = Inventory()
386
>>> inv.add(InventoryEntry('123123', 'hello.c', 'file', ROOT_ID))
1222
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
1223
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
387
1224
>>> inv['123123'].name
391
1228
return self._byid[file_id]
392
1229
except KeyError:
394
raise BzrError("can't look up file_id None")
396
raise BzrError("file_id {%s} not in inventory" % file_id)
1230
# really we're passing an inventory, not a tree...
1231
raise errors.NoSuchId(self, file_id)
399
1233
def get_file_kind(self, file_id):
400
1234
return self._byid[file_id].kind
520
1321
>>> i2 = Inventory()
523
>>> i1.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
1324
>>> i1.add(InventoryFile('123', 'foo', ROOT_ID))
1325
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
526
>>> i2.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
1328
>>> i2.add(InventoryFile('123', 'foo', ROOT_ID))
1329
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
530
1333
if not isinstance(other, Inventory):
531
1334
return NotImplemented
533
if len(self._byid) != len(other._byid):
534
# shortcut: obviously not the same
537
1336
return self._byid == other._byid
540
1338
def __ne__(self, other):
541
return not (self == other)
1339
return not self.__eq__(other)
544
1341
def __hash__(self):
545
1342
raise ValueError('not hashable')
549
def get_idpath(self, file_id):
550
"""Return a list of file_ids for the path to an entry.
552
The list contains one element for each directory followed by
553
the id of the file itself. So the length of the returned list
554
is equal to the depth of the file in the tree, counting the
555
root directory as depth 1.
558
while file_id != None:
1344
def _iter_file_id_parents(self, file_id):
1345
"""Yield the parents of file_id up to the root."""
1346
while file_id is not None:
560
1348
ie = self._byid[file_id]
561
1349
except KeyError:
562
raise BzrError("file_id {%s} not found in inventory" % file_id)
563
p.insert(0, ie.file_id)
1350
raise errors.NoSuchId(tree=None, file_id=file_id)
564
1352
file_id = ie.parent_id
568
def id2path(self, file_id):
569
"""Return as a list the path to file_id."""
571
# get all names, skipping root
572
p = [self._byid[fid].name for fid in self.get_idpath(file_id)[1:]]
573
return os.sep.join(p)
577
def path2id(self, name):
578
"""Walk down through directories to return entry of last component.
580
names may be either a list of path components, or a single
581
string, in which case it is automatically split.
583
This returns the entry of the last component in the path,
584
which may be either a file or a directory.
586
Returns None iff the path is not found.
1354
def has_filename(self, filename):
1355
return bool(self.path2id(filename))
1357
def has_id(self, file_id):
1358
return (file_id in self._byid)
1360
def _make_delta(self, old):
1361
"""Make an inventory delta from two inventories."""
1362
old_getter = getattr(old, '_byid', old)
1363
new_getter = self._byid
1364
old_ids = set(old_getter)
1365
new_ids = set(new_getter)
1366
adds = new_ids - old_ids
1367
deletes = old_ids - new_ids
1368
if not adds and not deletes:
1371
common = old_ids.intersection(new_ids)
1373
for file_id in deletes:
1374
delta.append((old.id2path(file_id), None, file_id, None))
1375
for file_id in adds:
1376
delta.append((None, self.id2path(file_id), file_id, self[file_id]))
1377
for file_id in common:
1378
new_ie = new_getter[file_id]
1379
old_ie = old_getter[file_id]
1380
# If xml_serializer returns the cached InventoryEntries (rather
1381
# than always doing .copy()), inlining the 'is' check saves 2.7M
1382
# calls to __eq__. Under lsprof this saves 20s => 6s.
1383
# It is a minor improvement without lsprof.
1384
if old_ie is new_ie or old_ie == new_ie:
1387
delta.append((old.id2path(file_id), self.id2path(file_id),
1391
def remove_recursive_id(self, file_id):
1392
"""Remove file_id, and children, from the inventory.
1394
:param file_id: A file_id to remove.
588
if isinstance(name, types.StringTypes):
589
name = splitpath(name)
591
mutter("lookup path %r" % name)
596
cie = parent.children[f]
598
assert cie.parent_id == parent.file_id
604
return parent.file_id
607
def has_filename(self, names):
608
return bool(self.path2id(names))
611
def has_id(self, file_id):
612
return self._byid.has_key(file_id)
1396
to_find_delete = [self._byid[file_id]]
1398
while to_find_delete:
1399
ie = to_find_delete.pop()
1400
to_delete.append(ie.file_id)
1401
if ie.kind == 'directory':
1402
to_find_delete.extend(ie.children.values())
1403
for file_id in reversed(to_delete):
1405
del self._byid[file_id]
1406
if ie.parent_id is not None:
1407
del self[ie.parent_id].children[ie.name]
615
1411
def rename(self, file_id, new_parent_id, new_name):
616
1412
"""Move a file within the inventory.
618
1414
This can change either the name, or the parent, or both.
620
This does not move the working file."""
1416
This does not move the working file.
1418
new_name = ensure_normalized_name(new_name)
621
1419
if not is_valid_name(new_name):
622
1420
raise BzrError("not an acceptable filename: %r" % new_name)
638
1436
del old_parent.children[file_ie.name]
639
1437
new_parent.children[new_name] = file_ie
641
1439
file_ie.name = new_name
642
1440
file_ie.parent_id = new_parent_id
1442
def is_root(self, file_id):
1443
return self.root is not None and file_id == self.root.file_id
1446
class CHKInventory(CommonInventory):
1447
"""An inventory persisted in a CHK store.
1449
By design, a CHKInventory is immutable so many of the methods
1450
supported by Inventory - add, rename, apply_delta, etc - are *not*
1451
supported. To create a new CHKInventory, use create_by_apply_delta()
1452
or from_inventory(), say.
1454
Internally, a CHKInventory has one or two CHKMaps:
1456
* id_to_entry - a map from (file_id,) => InventoryEntry as bytes
1457
* parent_id_basename_to_file_id - a map from (parent_id, basename_utf8)
1460
The second map is optional and not present in early CHkRepository's.
1462
No caching is performed: every method call or item access will perform
1463
requests to the storage layer. As such, keep references to objects you
1467
def __init__(self, search_key_name):
1468
CommonInventory.__init__(self)
1469
self._fileid_to_entry_cache = {}
1470
self._path_to_fileid_cache = {}
1471
self._search_key_name = search_key_name
1473
def _entry_to_bytes(self, entry):
1474
"""Serialise entry as a single bytestring.
1476
:param Entry: An inventory entry.
1477
:return: A bytestring for the entry.
1480
ENTRY ::= FILE | DIR | SYMLINK | TREE
1481
FILE ::= "file: " COMMON SEP SHA SEP SIZE SEP EXECUTABLE
1482
DIR ::= "dir: " COMMON
1483
SYMLINK ::= "symlink: " COMMON SEP TARGET_UTF8
1484
TREE ::= "tree: " COMMON REFERENCE_REVISION
1485
COMMON ::= FILE_ID SEP PARENT_ID SEP NAME_UTF8 SEP REVISION
1488
if entry.parent_id is not None:
1489
parent_str = entry.parent_id
1492
name_str = entry.name.encode("utf8")
1493
if entry.kind == 'file':
1494
if entry.executable:
1498
return "file: %s\n%s\n%s\n%s\n%s\n%d\n%s" % (
1499
entry.file_id, parent_str, name_str, entry.revision,
1500
entry.text_sha1, entry.text_size, exec_str)
1501
elif entry.kind == 'directory':
1502
return "dir: %s\n%s\n%s\n%s" % (
1503
entry.file_id, parent_str, name_str, entry.revision)
1504
elif entry.kind == 'symlink':
1505
return "symlink: %s\n%s\n%s\n%s\n%s" % (
1506
entry.file_id, parent_str, name_str, entry.revision,
1507
entry.symlink_target.encode("utf8"))
1508
elif entry.kind == 'tree-reference':
1509
return "tree: %s\n%s\n%s\n%s\n%s" % (
1510
entry.file_id, parent_str, name_str, entry.revision,
1511
entry.reference_revision)
1513
raise ValueError("unknown kind %r" % entry.kind)
1516
def _bytes_to_utf8name_key(bytes):
1517
"""Get the file_id, revision_id key out of bytes."""
1518
# We don't normally care about name, except for times when we want
1519
# to filter out empty names because of non rich-root...
1520
sections = bytes.split('\n')
1521
kind, file_id = sections[0].split(': ')
1522
return (sections[2], file_id, sections[3])
1524
def _bytes_to_entry(self, bytes):
1525
"""Deserialise a serialised entry."""
1526
sections = bytes.split('\n')
1527
if sections[0].startswith("file: "):
1528
result = InventoryFile(sections[0][6:],
1529
sections[2].decode('utf8'),
1531
result.text_sha1 = sections[4]
1532
result.text_size = int(sections[5])
1533
result.executable = sections[6] == "Y"
1534
elif sections[0].startswith("dir: "):
1535
result = CHKInventoryDirectory(sections[0][5:],
1536
sections[2].decode('utf8'),
1538
elif sections[0].startswith("symlink: "):
1539
result = InventoryLink(sections[0][9:],
1540
sections[2].decode('utf8'),
1542
result.symlink_target = sections[4].decode('utf8')
1543
elif sections[0].startswith("tree: "):
1544
result = TreeReference(sections[0][6:],
1545
sections[2].decode('utf8'),
1547
result.reference_revision = sections[4]
1549
raise ValueError("Not a serialised entry %r" % bytes)
1550
result.revision = sections[3]
1551
if result.parent_id == '':
1552
result.parent_id = None
1553
self._fileid_to_entry_cache[result.file_id] = result
1556
def _get_mutable_inventory(self):
1557
"""See CommonInventory._get_mutable_inventory."""
1558
entries = self.iter_entries()
1559
inv = Inventory(None, self.revision_id)
1560
for path, inv_entry in entries:
1561
inv.add(inv_entry.copy())
1564
def create_by_apply_delta(self, inventory_delta, new_revision_id,
1565
propagate_caches=False):
1566
"""Create a new CHKInventory by applying inventory_delta to this one.
1568
:param inventory_delta: The inventory delta to apply. See
1569
Inventory.apply_delta for details.
1570
:param new_revision_id: The revision id of the resulting CHKInventory.
1571
:param propagate_caches: If True, the caches for this inventory are
1572
copied to and updated for the result.
1573
:return: The new CHKInventory.
1575
result = CHKInventory(self._search_key_name)
1576
if propagate_caches:
1577
# Just propagate the path-to-fileid cache for now
1578
result._path_to_fileid_cache = dict(self._path_to_fileid_cache.iteritems())
1579
search_key_func = chk_map.search_key_registry.get(self._search_key_name)
1580
self.id_to_entry._ensure_root()
1581
maximum_size = self.id_to_entry._root_node.maximum_size
1582
result.revision_id = new_revision_id
1583
result.id_to_entry = chk_map.CHKMap(
1584
self.id_to_entry._store,
1585
self.id_to_entry.key(),
1586
search_key_func=search_key_func)
1587
result.id_to_entry._ensure_root()
1588
result.id_to_entry._root_node.set_maximum_size(maximum_size)
1589
parent_id_basename_delta = []
1590
if self.parent_id_basename_to_file_id is not None:
1591
result.parent_id_basename_to_file_id = chk_map.CHKMap(
1592
self.parent_id_basename_to_file_id._store,
1593
self.parent_id_basename_to_file_id.key(),
1594
search_key_func=search_key_func)
1595
result.parent_id_basename_to_file_id._ensure_root()
1596
self.parent_id_basename_to_file_id._ensure_root()
1597
result_p_id_root = result.parent_id_basename_to_file_id._root_node
1598
p_id_root = self.parent_id_basename_to_file_id._root_node
1599
result_p_id_root.set_maximum_size(p_id_root.maximum_size)
1600
result_p_id_root._key_width = p_id_root._key_width
1602
result.parent_id_basename_to_file_id = None
1603
result.root_id = self.root_id
1604
id_to_entry_delta = []
1605
for old_path, new_path, file_id, entry in inventory_delta:
1608
result.root_id = file_id
1609
if new_path is None:
1614
if propagate_caches:
1616
del result._path_to_fileid_cache[old_path]
1620
new_key = (file_id,)
1621
new_value = result._entry_to_bytes(entry)
1622
# Update caches. It's worth doing this whether
1623
# we're propagating the old caches or not.
1624
result._path_to_fileid_cache[new_path] = file_id
1625
if old_path is None:
1628
old_key = (file_id,)
1629
id_to_entry_delta.append((old_key, new_key, new_value))
1630
if result.parent_id_basename_to_file_id is not None:
1631
# parent_id, basename changes
1632
if old_path is None:
1635
old_entry = self[file_id]
1636
old_key = self._parent_id_basename_key(old_entry)
1637
if new_path is None:
1641
new_key = self._parent_id_basename_key(entry)
1643
if old_key != new_key:
1644
# If the two keys are the same, the value will be unchanged
1645
# as its always the file id.
1646
parent_id_basename_delta.append((old_key, new_key, new_value))
1647
result.id_to_entry.apply_delta(id_to_entry_delta)
1648
if parent_id_basename_delta:
1649
result.parent_id_basename_to_file_id.apply_delta(parent_id_basename_delta)
1653
def deserialise(klass, chk_store, bytes, expected_revision_id):
1654
"""Deserialise a CHKInventory.
1656
:param chk_store: A CHK capable VersionedFiles instance.
1657
:param bytes: The serialised bytes.
1658
:param expected_revision_id: The revision ID we think this inventory is
1660
:return: A CHKInventory
1662
lines = bytes.split('\n')
1664
raise AssertionError('bytes to deserialize must end with an eol')
1666
if lines[0] != 'chkinventory:':
1667
raise ValueError("not a serialised CHKInventory: %r" % bytes)
1669
allowed_keys = frozenset(['root_id', 'revision_id', 'search_key_name',
1670
'parent_id_basename_to_file_id',
1672
for line in lines[1:]:
1673
key, value = line.split(': ', 1)
1674
if key not in allowed_keys:
1675
raise errors.BzrError('Unknown key in inventory: %r\n%r'
1678
raise errors.BzrError('Duplicate key in inventory: %r\n%r'
1681
revision_id = info['revision_id']
1682
root_id = info['root_id']
1683
search_key_name = info.get('search_key_name', 'plain')
1684
parent_id_basename_to_file_id = info.get(
1685
'parent_id_basename_to_file_id', None)
1686
id_to_entry = info['id_to_entry']
1688
result = CHKInventory(search_key_name)
1689
result.revision_id = revision_id
1690
result.root_id = root_id
1691
search_key_func = chk_map.search_key_registry.get(
1692
result._search_key_name)
1693
if parent_id_basename_to_file_id is not None:
1694
result.parent_id_basename_to_file_id = chk_map.CHKMap(
1695
chk_store, (parent_id_basename_to_file_id,),
1696
search_key_func=search_key_func)
1698
result.parent_id_basename_to_file_id = None
1700
result.id_to_entry = chk_map.CHKMap(chk_store, (id_to_entry,),
1701
search_key_func=search_key_func)
1702
if (result.revision_id,) != expected_revision_id:
1703
raise ValueError("Mismatched revision id and expected: %r, %r" %
1704
(result.revision_id, expected_revision_id))
1708
def from_inventory(klass, chk_store, inventory, maximum_size=0, search_key_name='plain'):
1709
"""Create a CHKInventory from an existing inventory.
1711
The content of inventory is copied into the chk_store, and a
1712
CHKInventory referencing that is returned.
1714
:param chk_store: A CHK capable VersionedFiles instance.
1715
:param inventory: The inventory to copy.
1716
:param maximum_size: The CHKMap node size limit.
1717
:param search_key_name: The identifier for the search key function
1719
result = CHKInventory(search_key_name)
1720
result.revision_id = inventory.revision_id
1721
result.root_id = inventory.root.file_id
1722
search_key_func = chk_map.search_key_registry.get(search_key_name)
1723
result.id_to_entry = chk_map.CHKMap(chk_store, None, search_key_func)
1724
result.id_to_entry._root_node.set_maximum_size(maximum_size)
1726
result.parent_id_basename_to_file_id = chk_map.CHKMap(chk_store,
1727
None, search_key_func)
1728
result.parent_id_basename_to_file_id._root_node.set_maximum_size(
1730
result.parent_id_basename_to_file_id._root_node._key_width = 2
1731
parent_id_delta = []
1732
for path, entry in inventory.iter_entries():
1733
file_id_delta.append((None, (entry.file_id,),
1734
result._entry_to_bytes(entry)))
1735
parent_id_delta.append(
1736
(None, result._parent_id_basename_key(entry),
1738
result.id_to_entry.apply_delta(file_id_delta)
1739
result.parent_id_basename_to_file_id.apply_delta(parent_id_delta)
1742
def _parent_id_basename_key(self, entry):
1743
"""Create a key for a entry in a parent_id_basename_to_file_id index."""
1744
if entry.parent_id is not None:
1745
parent_id = entry.parent_id
1748
return parent_id, entry.name.encode('utf8')
1750
def __getitem__(self, file_id):
1751
"""map a single file_id -> InventoryEntry."""
1753
raise errors.NoSuchId(self, file_id)
1754
result = self._fileid_to_entry_cache.get(file_id, None)
1755
if result is not None:
1758
return self._bytes_to_entry(
1759
self.id_to_entry.iteritems([(file_id,)]).next()[1])
1760
except StopIteration:
1761
# really we're passing an inventory, not a tree...
1762
raise errors.NoSuchId(self, file_id)
1764
def has_id(self, file_id):
1765
# Perhaps have an explicit 'contains' method on CHKMap ?
1766
if self._fileid_to_entry_cache.get(file_id, None) is not None:
1768
return len(list(self.id_to_entry.iteritems([(file_id,)]))) == 1
1770
def is_root(self, file_id):
1771
return file_id == self.root_id
1773
def _iter_file_id_parents(self, file_id):
1774
"""Yield the parents of file_id up to the root."""
1775
while file_id is not None:
1779
raise errors.NoSuchId(tree=self, file_id=file_id)
1781
file_id = ie.parent_id
1784
"""Iterate over all file-ids."""
1785
for key, _ in self.id_to_entry.iteritems():
1788
def iter_just_entries(self):
1789
"""Iterate over all entries.
1791
Unlike iter_entries(), just the entries are returned (not (path, ie))
1792
and the order of entries is undefined.
1794
XXX: We may not want to merge this into bzr.dev.
1796
for key, entry in self.id_to_entry.iteritems():
1798
ie = self._fileid_to_entry_cache.get(file_id, None)
1800
ie = self._bytes_to_entry(entry)
1801
self._fileid_to_entry_cache[file_id] = ie
1804
def iter_changes(self, basis):
1805
"""Generate a Tree.iter_changes change list between this and basis.
1807
:param basis: Another CHKInventory.
1808
:return: An iterator over the changes between self and basis, as per
1809
tree.iter_changes().
1811
# We want: (file_id, (path_in_source, path_in_target),
1812
# changed_content, versioned, parent, name, kind,
1814
for key, basis_value, self_value in \
1815
self.id_to_entry.iter_changes(basis.id_to_entry):
1817
if basis_value is not None:
1818
basis_entry = basis._bytes_to_entry(basis_value)
1819
path_in_source = basis.id2path(file_id)
1820
basis_parent = basis_entry.parent_id
1821
basis_name = basis_entry.name
1822
basis_executable = basis_entry.executable
1824
path_in_source = None
1827
basis_executable = None
1828
if self_value is not None:
1829
self_entry = self._bytes_to_entry(self_value)
1830
path_in_target = self.id2path(file_id)
1831
self_parent = self_entry.parent_id
1832
self_name = self_entry.name
1833
self_executable = self_entry.executable
1835
path_in_target = None
1838
self_executable = None
1839
if basis_value is None:
1841
kind = (None, self_entry.kind)
1842
versioned = (False, True)
1843
elif self_value is None:
1845
kind = (basis_entry.kind, None)
1846
versioned = (True, False)
1848
kind = (basis_entry.kind, self_entry.kind)
1849
versioned = (True, True)
1850
changed_content = False
1851
if kind[0] != kind[1]:
1852
changed_content = True
1853
elif kind[0] == 'file':
1854
if (self_entry.text_size != basis_entry.text_size or
1855
self_entry.text_sha1 != basis_entry.text_sha1):
1856
changed_content = True
1857
elif kind[0] == 'symlink':
1858
if self_entry.symlink_target != basis_entry.symlink_target:
1859
changed_content = True
1860
elif kind[0] == 'tree-reference':
1861
if (self_entry.reference_revision !=
1862
basis_entry.reference_revision):
1863
changed_content = True
1864
parent = (basis_parent, self_parent)
1865
name = (basis_name, self_name)
1866
executable = (basis_executable, self_executable)
1867
if (not changed_content
1868
and parent[0] == parent[1]
1869
and name[0] == name[1]
1870
and executable[0] == executable[1]):
1871
# Could happen when only the revision changed for a directory
1874
yield (file_id, (path_in_source, path_in_target), changed_content,
1875
versioned, parent, name, kind, executable)
1878
"""Return the number of entries in the inventory."""
1879
return len(self.id_to_entry)
1881
def _make_delta(self, old):
1882
"""Make an inventory delta from two inventories."""
1883
if type(old) != CHKInventory:
1884
return CommonInventory._make_delta(self, old)
1886
for key, old_value, self_value in \
1887
self.id_to_entry.iter_changes(old.id_to_entry):
1889
if old_value is not None:
1890
old_path = old.id2path(file_id)
1893
if self_value is not None:
1894
entry = self._bytes_to_entry(self_value)
1895
self._fileid_to_entry_cache[file_id] = entry
1896
new_path = self.id2path(file_id)
1900
delta.append((old_path, new_path, file_id, entry))
1903
def path2id(self, name):
1904
"""See CommonInventory.path2id()."""
1905
result = self._path_to_fileid_cache.get(name, None)
1907
result = CommonInventory.path2id(self, name)
1908
self._path_to_fileid_cache[name] = result
1912
"""Serialise the inventory to lines."""
1913
lines = ["chkinventory:\n"]
1914
if self._search_key_name != 'plain':
1915
# custom ordering grouping things that don't change together
1916
lines.append('search_key_name: %s\n' % (self._search_key_name,))
1917
lines.append("root_id: %s\n" % self.root_id)
1918
lines.append('parent_id_basename_to_file_id: %s\n' %
1919
self.parent_id_basename_to_file_id.key())
1920
lines.append("revision_id: %s\n" % self.revision_id)
1921
lines.append("id_to_entry: %s\n" % self.id_to_entry.key())
1923
lines.append("revision_id: %s\n" % self.revision_id)
1924
lines.append("root_id: %s\n" % self.root_id)
1925
if self.parent_id_basename_to_file_id is not None:
1926
lines.append('parent_id_basename_to_file_id: %s\n' %
1927
self.parent_id_basename_to_file_id.key())
1928
lines.append("id_to_entry: %s\n" % self.id_to_entry.key())
1933
"""Get the root entry."""
1934
return self[self.root_id]
1937
class CHKInventoryDirectory(InventoryDirectory):
1938
"""A directory in an inventory."""
1940
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
1941
'text_id', 'parent_id', '_children', 'executable',
1942
'revision', 'symlink_target', 'reference_revision',
1945
def __init__(self, file_id, name, parent_id, chk_inventory):
1946
# Don't call InventoryDirectory.__init__ - it isn't right for this
1948
InventoryEntry.__init__(self, file_id, name, parent_id)
1949
self._children = None
1950
self.kind = 'directory'
1951
self._chk_inventory = chk_inventory
1955
"""Access the list of children of this directory.
1957
With a parent_id_basename_to_file_id index, loads all the children,
1958
without loads the entire index. Without is bad. A more sophisticated
1959
proxy object might be nice, to allow partial loading of children as
1960
well when specific names are accessed. (So path traversal can be
1961
written in the obvious way but not examine siblings.).
1963
if self._children is not None:
1964
return self._children
1965
# No longer supported
1966
if self._chk_inventory.parent_id_basename_to_file_id is None:
1967
raise AssertionError("Inventories without"
1968
" parent_id_basename_to_file_id are no longer supported")
1970
# XXX: Todo - use proxy objects for the children rather than loading
1971
# all when the attribute is referenced.
1972
parent_id_index = self._chk_inventory.parent_id_basename_to_file_id
1974
for (parent_id, name_utf8), file_id in parent_id_index.iteritems(
1975
key_filter=[(self.file_id,)]):
1976
child_keys.add((file_id,))
1978
for file_id_key in child_keys:
1979
entry = self._chk_inventory._fileid_to_entry_cache.get(
1980
file_id_key[0], None)
1981
if entry is not None:
1982
result[entry.name] = entry
1983
cached.add(file_id_key)
1984
child_keys.difference_update(cached)
1985
# populate; todo: do by name
1986
id_to_entry = self._chk_inventory.id_to_entry
1987
for file_id_key, bytes in id_to_entry.iteritems(child_keys):
1988
entry = self._chk_inventory._bytes_to_entry(bytes)
1989
result[entry.name] = entry
1990
self._chk_inventory._fileid_to_entry_cache[file_id_key[0]] = entry
1991
self._children = result
1995
'directory': InventoryDirectory,
1996
'file': InventoryFile,
1997
'symlink': InventoryLink,
1998
'tree-reference': TreeReference
2001
def make_entry(kind, name, parent_id, file_id=None):
2002
"""Create an inventory entry.
2004
:param kind: the type of inventory entry to create.
2005
:param name: the basename of the entry.
2006
:param parent_id: the parent_id of the entry.
2007
:param file_id: the file_id to use. if None, one will be created.
2010
file_id = generate_ids.gen_file_id(name)
2011
name = ensure_normalized_name(name)
2013
factory = entry_factory[kind]
2015
raise BzrError("unknown kind %r" % kind)
2016
return factory(file_id, name, parent_id)
2019
def ensure_normalized_name(name):
2022
:raises InvalidNormalization: When name is not normalized, and cannot be
2023
accessed on this platform by the normalized path.
2024
:return: The NFC normalised version of name.
2026
#------- This has been copied to bzrlib.dirstate.DirState.add, please
2027
# keep them synchronised.
2028
# we dont import normalized_filename directly because we want to be
2029
# able to change the implementation at runtime for tests.
2030
norm_name, can_access = osutils.normalized_filename(name)
2031
if norm_name != name:
2035
# TODO: jam 20060701 This would probably be more useful
2036
# if the error was raised with the full path
2037
raise errors.InvalidNormalization(name)
649
2043
def is_valid_name(name):
2045
if _NAME_RE is None:
652
2046
_NAME_RE = re.compile(r'^[^/\\]+$')
654
2048
return bool(_NAME_RE.match(name))
added
removed
bzrlib/inventory.py
