36
59
An InventoryEntry has the following fields, which are also
37
60
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.
65
(within the parent directory)
68
file_id of the parent directory, or ROOT_ID
71
the revision_id in which this variation of this file was
75
Indicates that this file should be executable on systems
79
sha-1 of the text of the file
82
size in bytes of the text of the file
84
(reading a version 4 tree created a text_id field.)
54
86
>>> 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():
89
>>> i.add(InventoryDirectory('123', 'src', ROOT_ID))
90
InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None)
91
>>> 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])
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):
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'))
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))
100
>>> i.add(InventoryFile('2324', 'bye.c', '123'))
101
InventoryFile('2324', 'bye.c', parent_id='123', sha1=None, len=None)
102
>>> i.add(InventoryDirectory('2325', 'wibble', '123'))
103
InventoryDirectory('2325', 'wibble', parent_id='123', revision=None)
70
104
>>> i.path2id('src/wibble')
74
>>> i.add(InventoryEntry('2326', 'wibble.c', 'file', '2325'))
108
>>> i.add(InventoryFile('2326', 'wibble.c', '2325'))
109
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
76
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
77
>>> for j in i.iter_entries():
79
... assert i.path2id(j[0])
111
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
112
>>> for path, entry in i.iter_entries():
114
... assert i.path2id(path)
85
121
src/wibble/wibble.c
86
122
>>> i.id2path('2326')
87
123
'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):
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'
136
def detect_changes(self, old_entry):
137
"""Return a (text_modified, meta_modified) from this to old_entry.
139
_read_tree_state must have been called on self and old_entry prior to
140
calling detect_changes.
144
def diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
145
output_to, reverse=False):
146
"""Perform a diff from this to to_entry.
148
text_diff will be used for textual difference calculation.
149
This is a template method, override _diff in child classes.
151
self._read_tree_state(tree.id2path(self.file_id), tree)
153
# cannot diff from one kind to another - you must do a removal
154
# and an addif they do not match.
155
assert self.kind == to_entry.kind
156
to_entry._read_tree_state(to_tree.id2path(to_entry.file_id),
158
self._diff(text_diff, from_label, tree, to_label, to_entry, to_tree,
161
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
162
output_to, reverse=False):
163
"""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.
171
Returned as a map from revision to inventory entry.
173
This is a map containing the file revisions in all parents
174
for which the file exists, and its revision is not a parent of
175
any other. If the file is new, the set will be empty.
177
:param versioned_file_store: A store where ancestry data on this
178
file id can be queried.
179
:param transaction: The transaction that queries to the versioned
180
file store should be completed under.
181
:param entry_vf: The entry versioned file, if its already available.
183
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.
189
# revision: ancestor list for each head
191
# identify candidate head revision ids.
192
for inv in previous_inventories:
193
if self.file_id in inv:
194
ie = inv[self.file_id]
195
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
205
if candidates[ie.revision].executable != ie.executable:
206
candidates[ie.revision].executable = False
207
ie.executable = False
208
except AttributeError:
210
# must now be the same.
211
assert candidates[ie.revision] == ie
213
# add this revision as a candidate.
214
candidates[ie.revision] = ie
216
# common case optimisation
217
if len(candidates) == 1:
218
# if there is only one candidate revision found
219
# then we can opening the versioned file to access ancestry:
220
# there cannot be any ancestors to eliminate when there is
221
# only one revision available.
222
heads[ie.revision] = ie
225
# eliminate ancestors amongst the available candidates:
226
# heads are those that are not an ancestor of any other candidate
227
# - this provides convergence at a per-file level.
228
for ie in candidates.values():
229
# may be an ancestor of a known head:
230
already_present = 0 != len(
231
[head for head in heads
232
if ie.revision in head_ancestors[head]])
234
# an ancestor of an analyzed candidate.
236
# not an ancestor of a known head:
237
# load the versioned file for this file id if needed
239
entry_vf = versioned_file_store.get_weave_or_empty(
240
self.file_id, transaction)
241
ancestors = get_ancestors(entry_vf, ie)
242
# may knock something else out:
243
check_heads = list(heads.keys())
244
for head in check_heads:
245
if head in ancestors:
246
# this previously discovered 'head' is not
247
# really a head - its an ancestor of the newly
250
head_ancestors[ie.revision] = ancestors
251
heads[ie.revision] = ie
254
def get_tar_item(self, root, dp, now, tree):
255
"""Get a tarfile item and a file stream for its content."""
256
item = tarfile.TarInfo(osutils.pathjoin(root, dp).encode('utf8'))
257
# TODO: would be cool to actually set it to the timestamp of the
258
# revision it was last changed
260
fileobj = self._put_in_tar(item, tree)
264
"""Return true if the object this entry represents has textual data.
266
Note that textual data includes binary content.
268
Also note that all entries get weave files created for them.
269
This attribute is primarily used when upgrading from old trees that
270
did not have the weave index for all inventory entries.
274
def __init__(self, file_id, name, parent_id, text_id=None):
101
275
"""Create an InventoryEntry
103
277
The filename must be a single component, relative to the
104
278
parent directory; it cannot be a whole path or relative name.
106
>>> e = InventoryEntry('123', 'hello.c', 'file', ROOT_ID)
280
>>> e = InventoryFile('123', 'hello.c', ROOT_ID)
111
>>> e = InventoryEntry('123', 'src/hello.c', 'file', ROOT_ID)
285
>>> e = InventoryFile('123', 'src/hello.c', ROOT_ID)
112
286
Traceback (most recent call last):
113
BzrCheckError: InventoryEntry name 'src/hello.c' is invalid
287
InvalidEntryName: Invalid entry name: src/hello.c
289
assert isinstance(name, basestring), name
115
290
if '/' in name or '\\' in name:
116
raise BzrCheckError('InventoryEntry name %r is invalid' % name)
291
raise errors.InvalidEntryName(name=name)
292
self.executable = False
118
294
self.text_sha1 = None
119
295
self.text_size = None
121
296
self.file_id = file_id
297
assert isinstance(file_id, (str, None.__class__)), \
298
'bad type %r for %r' % (type(file_id), file_id)
124
300
self.text_id = text_id
125
301
self.parent_id = parent_id
126
if kind == 'directory':
131
raise BzrError("unhandled entry kind %r" % kind)
302
self.symlink_target = None
303
self.reference_revision = None
305
def kind_character(self):
306
"""Return a short kind indicator useful for appending to names."""
307
raise BzrError('unknown kind %r' % self.kind)
309
known_kinds = ('file', 'directory', 'symlink')
311
def _put_in_tar(self, item, tree):
312
"""populate item for stashing in a tar, and return the content stream.
314
If no content is available, return None.
316
raise BzrError("don't know how to export {%s} of kind %r" %
317
(self.file_id, self.kind))
319
def put_on_disk(self, dest, dp, tree):
320
"""Create a representation of self on disk in the prefix dest.
322
This is a template method - implement _put_on_disk in subclasses.
324
fullpath = osutils.pathjoin(dest, dp)
325
self._put_on_disk(fullpath, tree)
326
# mutter(" export {%s} kind %s to %s", self.file_id,
327
# self.kind, fullpath)
329
def _put_on_disk(self, fullpath, tree):
330
"""Put this entry onto disk at fullpath, from tree tree."""
331
raise BzrError("don't know how to export {%s} of kind %r" % (self.file_id, self.kind))
135
333
def sorted_children(self):
136
l = self.children.items()
334
return sorted(self.children.items())
337
def versionable_kind(kind):
338
return (kind in ('file', 'directory', 'symlink', 'tree-reference'))
340
def check(self, checker, rev_id, inv, tree):
341
"""Check this inventory entry is intact.
343
This is a template method, override _check for kind specific
346
:param checker: Check object providing context for the checks;
347
can be used to find out what parts of the repository have already
349
:param rev_id: Revision id from which this InventoryEntry was loaded.
350
Not necessarily the last-changed revision for this file.
351
:param inv: Inventory from which the entry was loaded.
352
:param tree: RevisionTree for this entry.
354
if self.parent_id is not None:
355
if not inv.has_id(self.parent_id):
356
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
357
% (self.parent_id, rev_id))
358
self._check(checker, rev_id, tree)
360
def _check(self, checker, rev_id, tree):
361
"""Check this inventory entry for kind specific errors."""
362
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
366
"""Clone this inventory entry."""
367
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:
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
151
411
def __repr__(self):
152
return ("%s(%r, %r, kind=%r, parent_id=%r)"
412
return ("%s(%r, %r, parent_id=%r, revision=%r)"
153
413
% (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)
419
def snapshot(self, revision, path, previous_entries,
420
work_tree, commit_builder):
421
"""Make a snapshot of this entry which may or may not have changed.
423
This means that all its fields are populated, that it has its
424
text stored in the text store or weave.
426
:return: True if anything was recorded
428
# cannot be unchanged unless there is only one parent file rev.
429
self._read_tree_state(path, work_tree)
430
if len(previous_entries) == 1:
431
parent_ie = previous_entries.values()[0]
432
if self._unchanged(parent_ie):
433
self.revision = parent_ie.revision
435
self.revision = revision
436
return self._snapshot_text(previous_entries, work_tree, commit_builder)
438
def _snapshot_text(self, file_parents, work_tree, commit_builder):
439
"""Record the 'text' of this entry, whatever form that takes.
441
:return: True if anything was recorded
443
raise NotImplementedError(self._snapshot_text)
214
445
def __eq__(self, other):
215
446
if not isinstance(other, InventoryEntry):
216
447
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)
449
return ((self.file_id == other.file_id)
450
and (self.name == other.name)
451
and (other.symlink_target == self.symlink_target)
452
and (self.text_sha1 == other.text_sha1)
453
and (self.text_size == other.text_size)
454
and (self.text_id == other.text_id)
455
and (self.parent_id == other.parent_id)
456
and (self.kind == other.kind)
457
and (self.revision == other.revision)
458
and (self.executable == other.executable)
459
and (self.reference_revision == other.reference_revision)
227
462
def __ne__(self, other):
228
463
return not (self == other)
248
524
and (self.children == other.children)
527
class InventoryDirectory(InventoryEntry):
528
"""A directory in an inventory."""
530
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
531
'text_id', 'parent_id', 'children', 'executable',
532
'revision', 'symlink_target', 'reference_revision']
534
def _check(self, checker, rev_id, tree):
535
"""See InventoryEntry._check"""
536
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
537
raise BzrCheckError('directory {%s} has text in revision {%s}'
538
% (self.file_id, rev_id))
541
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
542
other.revision = self.revision
543
# note that children are *not* copied; they're pulled across when
547
def __init__(self, file_id, name, parent_id):
548
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
550
self.kind = 'directory'
552
def kind_character(self):
553
"""See InventoryEntry.kind_character."""
556
def _put_in_tar(self, item, tree):
557
"""See InventoryEntry._put_in_tar."""
558
item.type = tarfile.DIRTYPE
565
def _put_on_disk(self, fullpath, tree):
566
"""See InventoryEntry._put_on_disk."""
569
def _snapshot_text(self, file_parents, work_tree, commit_builder):
570
"""See InventoryEntry._snapshot_text."""
571
commit_builder.modified_directory(self.file_id, file_parents)
575
class InventoryFile(InventoryEntry):
576
"""A file in an inventory."""
578
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
579
'text_id', 'parent_id', 'children', 'executable',
580
'revision', 'symlink_target', 'reference_revision']
582
def _check(self, checker, tree_revision_id, tree):
583
"""See InventoryEntry._check"""
584
t = (self.file_id, self.revision)
585
if t in checker.checked_texts:
586
prev_sha = checker.checked_texts[t]
587
if prev_sha != self.text_sha1:
588
raise BzrCheckError('mismatched sha1 on {%s} in {%s}' %
589
(self.file_id, tree_revision_id))
591
checker.repeated_text_cnt += 1
594
if self.file_id not in checker.checked_weaves:
595
mutter('check weave {%s}', self.file_id)
596
w = tree._get_weave(self.file_id)
597
# Not passing a progress bar, because it creates a new
598
# progress, which overwrites the current progress,
599
# and doesn't look nice
601
checker.checked_weaves[self.file_id] = True
603
w = tree._get_weave(self.file_id)
605
mutter('check version {%s} of {%s}', tree_revision_id, self.file_id)
606
checker.checked_text_cnt += 1
607
# We can't check the length, because Weave doesn't store that
608
# information, and the whole point of looking at the weave's
609
# sha1sum is that we don't have to extract the text.
610
if self.text_sha1 != w.get_sha1(self.revision):
611
raise BzrCheckError('text {%s} version {%s} wrong sha1'
612
% (self.file_id, self.revision))
613
checker.checked_texts[t] = self.text_sha1
616
other = InventoryFile(self.file_id, self.name, self.parent_id)
617
other.executable = self.executable
618
other.text_id = self.text_id
619
other.text_sha1 = self.text_sha1
620
other.text_size = self.text_size
621
other.revision = self.revision
624
def detect_changes(self, old_entry):
625
"""See InventoryEntry.detect_changes."""
626
assert self.text_sha1 is not None
627
assert old_entry.text_sha1 is not None
628
text_modified = (self.text_sha1 != old_entry.text_sha1)
629
meta_modified = (self.executable != old_entry.executable)
630
return text_modified, meta_modified
632
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
633
output_to, reverse=False):
634
"""See InventoryEntry._diff."""
636
from_text = tree.get_file(self.file_id).readlines()
638
to_text = to_tree.get_file(to_entry.file_id).readlines()
642
text_diff(from_label, from_text,
643
to_label, to_text, output_to)
645
text_diff(to_label, to_text,
646
from_label, from_text, output_to)
647
except errors.BinaryFile:
649
label_pair = (to_label, from_label)
651
label_pair = (from_label, to_label)
652
print >> output_to, \
653
("Binary files %s and %s differ" % label_pair).encode('utf8')
656
"""See InventoryEntry.has_text."""
659
def __init__(self, file_id, name, parent_id):
660
super(InventoryFile, self).__init__(file_id, name, parent_id)
663
def kind_character(self):
664
"""See InventoryEntry.kind_character."""
667
def _put_in_tar(self, item, tree):
668
"""See InventoryEntry._put_in_tar."""
669
item.type = tarfile.REGTYPE
670
fileobj = tree.get_file(self.file_id)
671
item.size = self.text_size
672
if tree.is_executable(self.file_id):
678
def _put_on_disk(self, fullpath, tree):
679
"""See InventoryEntry._put_on_disk."""
680
osutils.pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
681
if tree.is_executable(self.file_id):
682
os.chmod(fullpath, 0755)
684
def _read_tree_state(self, path, work_tree):
685
"""See InventoryEntry._read_tree_state."""
686
self.text_sha1 = work_tree.get_file_sha1(self.file_id, path=path)
687
# FIXME: 20050930 probe for the text size when getting sha1
688
# in _read_tree_state
689
self.executable = work_tree.is_executable(self.file_id, path=path)
692
return ("%s(%r, %r, parent_id=%r, sha1=%r, len=%s)"
693
% (self.__class__.__name__,
700
def _forget_tree_state(self):
701
self.text_sha1 = None
703
def snapshot(self, revision, path, previous_entries,
704
work_tree, commit_builder):
705
"""See InventoryEntry.snapshot."""
706
# Note: We use a custom implementation of this method for files
707
# because it's a performance critical part of commit.
709
# If this is the initial commit for this file, we know the sha is
710
# coming later so skip calculating it now (in _read_tree_state())
711
if len(previous_entries) == 0:
712
self.executable = work_tree.is_executable(self.file_id, path=path)
714
self._read_tree_state(path, work_tree)
716
# If nothing is changed from the sole parent, there's nothing to do
717
if len(previous_entries) == 1:
718
parent_ie = previous_entries.values()[0]
719
if self._unchanged(parent_ie):
720
self.revision = parent_ie.revision
723
# Add the file to the repository
724
self.revision = revision
725
def get_content_byte_lines():
726
return work_tree.get_file(self.file_id, path).readlines()
727
self.text_sha1, self.text_size = commit_builder.modified_file_text(
728
self.file_id, previous_entries, get_content_byte_lines,
729
self.text_sha1, self.text_size)
732
def _unchanged(self, previous_ie):
733
"""See InventoryEntry._unchanged."""
734
compatible = super(InventoryFile, self)._unchanged(previous_ie)
735
if self.text_sha1 != previous_ie.text_sha1:
738
# FIXME: 20050930 probe for the text size when getting sha1
739
# in _read_tree_state
740
self.text_size = previous_ie.text_size
741
if self.executable != previous_ie.executable:
746
class InventoryLink(InventoryEntry):
747
"""A file in an inventory."""
749
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
750
'text_id', 'parent_id', 'children', 'executable',
751
'revision', 'symlink_target', 'reference_revision']
753
def _check(self, checker, rev_id, tree):
754
"""See InventoryEntry._check"""
755
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
756
raise BzrCheckError('symlink {%s} has text in revision {%s}'
757
% (self.file_id, rev_id))
758
if self.symlink_target is None:
759
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
760
% (self.file_id, rev_id))
763
other = InventoryLink(self.file_id, self.name, self.parent_id)
764
other.symlink_target = self.symlink_target
765
other.revision = self.revision
768
def detect_changes(self, old_entry):
769
"""See InventoryEntry.detect_changes."""
770
# FIXME: which _modified field should we use ? RBC 20051003
771
text_modified = (self.symlink_target != old_entry.symlink_target)
773
mutter(" symlink target changed")
774
meta_modified = False
775
return text_modified, meta_modified
777
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
778
output_to, reverse=False):
779
"""See InventoryEntry._diff."""
780
from_text = self.symlink_target
781
if to_entry is not None:
782
to_text = to_entry.symlink_target
787
print >>output_to, '=== target changed %r => %r' % (from_text, to_text)
790
print >>output_to, '=== target was %r' % self.symlink_target
792
print >>output_to, '=== target is %r' % self.symlink_target
794
def __init__(self, file_id, name, parent_id):
795
super(InventoryLink, self).__init__(file_id, name, parent_id)
796
self.kind = 'symlink'
798
def kind_character(self):
799
"""See InventoryEntry.kind_character."""
802
def _put_in_tar(self, item, tree):
803
"""See InventoryEntry._put_in_tar."""
804
item.type = tarfile.SYMTYPE
808
item.linkname = self.symlink_target
811
def _put_on_disk(self, fullpath, tree):
812
"""See InventoryEntry._put_on_disk."""
814
os.symlink(self.symlink_target, fullpath)
816
raise BzrError("Failed to create symlink %r -> %r, error: %s" % (fullpath, self.symlink_target, e))
818
def _read_tree_state(self, path, work_tree):
819
"""See InventoryEntry._read_tree_state."""
820
self.symlink_target = work_tree.get_symlink_target(self.file_id)
822
def _forget_tree_state(self):
823
self.symlink_target = None
825
def _unchanged(self, previous_ie):
826
"""See InventoryEntry._unchanged."""
827
compatible = super(InventoryLink, self)._unchanged(previous_ie)
828
if self.symlink_target != previous_ie.symlink_target:
832
def _snapshot_text(self, file_parents, work_tree, commit_builder):
833
"""See InventoryEntry._snapshot_text."""
834
commit_builder.modified_link(
835
self.file_id, file_parents, self.symlink_target)
839
class TreeReference(InventoryEntry):
841
kind = 'tree-reference'
843
def __init__(self, file_id, name, parent_id, revision=None,
844
reference_revision=None):
845
InventoryEntry.__init__(self, file_id, name, parent_id)
846
self.revision = revision
847
self.reference_revision = reference_revision
850
return TreeReference(self.file_id, self.name, self.parent_id,
851
self.revision, self.reference_revision)
853
def _snapshot_text(self, file_parents, work_tree, commit_builder):
854
commit_builder.modified_reference(self.file_id, file_parents)
857
def _read_tree_state(self, path, work_tree):
858
"""Populate fields in the inventory entry from the given tree.
860
self.reference_revision = work_tree.get_reference_revision(
863
def _forget_tree_state(self):
864
self.reference_revision = None
252
867
class Inventory(object):
253
868
"""Inventory of versioned files in a tree.
295
915
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)
918
if root_id is not None:
919
assert root_id.__class__ == str
920
self._set_root(InventoryDirectory(root_id, u'', None))
924
self.revision_id = revision_id
926
def _set_root(self, ie):
303
928
self._byid = {self.root.file_id: self.root}
931
# TODO: jam 20051218 Should copy also copy the revision_id?
932
entries = self.iter_entries()
933
other = Inventory(entries.next()[1].file_id)
934
# copy recursively so we know directories will be added before
935
# their children. There are more efficient ways than this...
936
for path, entry in entries():
937
other.add(entry.copy())
306
940
def __iter__(self):
307
941
return iter(self._byid)
310
943
def __len__(self):
311
944
"""Returns number of entries."""
312
945
return len(self._byid)
315
947
def iter_entries(self, from_dir=None):
316
948
"""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
950
if self.root is None:
954
elif isinstance(from_dir, basestring):
955
from_dir = self._byid[from_dir]
957
# unrolling the recursive called changed the time from
958
# 440ms/663ms (inline/total) to 116ms/116ms
959
children = from_dir.children.items()
961
children = collections.deque(children)
962
stack = [(u'', children)]
964
from_dir_relpath, children = stack[-1]
967
name, ie = children.popleft()
969
# we know that from_dir_relpath never ends in a slash
970
# and 'f' doesn't begin with one, we can do a string op, rather
971
# than the checks of pathjoin(), though this means that all paths
973
path = from_dir_relpath + '/' + name
977
if ie.kind != 'directory':
980
# But do this child first
981
new_children = ie.children.items()
983
new_children = collections.deque(new_children)
984
stack.append((path, new_children))
985
# Break out of inner loop, so that we start outer loop with child
988
# if we finished all children, pop it off the stack
991
def iter_entries_by_dir(self, from_dir=None, specific_file_ids=None):
992
"""Iterate over the entries in a directory first order.
994
This returns all entries for a directory before returning
995
the entries for children of a directory. This is not
996
lexicographically sorted order, and is a hybrid between
997
depth-first and breadth-first.
999
:return: This yields (path, entry) pairs
1001
if specific_file_ids:
1002
safe = osutils.safe_file_id
1003
specific_file_ids = set(safe(fid) for fid in specific_file_ids)
1004
# TODO? Perhaps this should return the from_dir so that the root is
1005
# yielded? or maybe an option?
1006
if from_dir is None:
1007
if self.root is None:
1009
# Optimize a common case
1010
if specific_file_ids is not None and len(specific_file_ids) == 1:
1011
file_id = list(specific_file_ids)[0]
1013
yield self.id2path(file_id), self[file_id]
1015
from_dir = self.root
1016
if (specific_file_ids is None or
1017
self.root.file_id in specific_file_ids):
1018
yield u'', self.root
1019
elif isinstance(from_dir, basestring):
1020
from_dir = self._byid[from_dir]
1022
if specific_file_ids is not None:
1023
# TODO: jam 20070302 This could really be done as a loop rather
1024
# than a bunch of recursive calls.
1027
def add_ancestors(file_id):
1028
if file_id not in byid:
1030
parent_id = byid[file_id].parent_id
1031
if parent_id is None:
1033
if parent_id not in parents:
1034
parents.add(parent_id)
1035
add_ancestors(parent_id)
1036
for file_id in specific_file_ids:
1037
add_ancestors(file_id)
1041
stack = [(u'', from_dir)]
1043
cur_relpath, cur_dir = stack.pop()
1046
for child_name, child_ie in sorted(cur_dir.children.iteritems()):
1048
child_relpath = cur_relpath + child_name
1050
if (specific_file_ids is None or
1051
child_ie.file_id in specific_file_ids):
1052
yield child_relpath, child_ie
1054
if child_ie.kind == 'directory':
1055
if parents is None or child_ie.file_id in parents:
1056
child_dirs.append((child_relpath+'/', child_ie))
1057
stack.extend(reversed(child_dirs))
1059
def make_entry(self, kind, name, parent_id, file_id=None):
1060
"""Simple thunk to bzrlib.inventory.make_entry."""
1061
return make_entry(kind, name, parent_id, file_id)
332
1063
def entries(self):
333
1064
"""Return list of (path, ie) for all entries except the root.
361
1091
for name, child_ie in kids:
362
child_path = os.path.join(parent_path, name)
1092
child_path = osutils.pathjoin(parent_path, name)
363
1093
descend(child_ie, child_path)
364
descend(self.root, '')
1094
descend(self.root, u'')
369
1097
def __contains__(self, file_id):
370
1098
"""True if this entry contains a file with given id.
372
1100
>>> inv = Inventory()
373
>>> inv.add(InventoryEntry('123', 'foo.c', 'file', ROOT_ID))
1101
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1102
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
374
1103
>>> '123' in inv
376
1105
>>> '456' in inv
379
return file_id in self._byid
1108
file_id = osutils.safe_file_id(file_id)
1109
return (file_id in self._byid)
382
1111
def __getitem__(self, file_id):
383
1112
"""Return the entry for given file_id.
385
1114
>>> inv = Inventory()
386
>>> inv.add(InventoryEntry('123123', 'hello.c', 'file', ROOT_ID))
1115
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
1116
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
387
1117
>>> inv['123123'].name
1120
file_id = osutils.safe_file_id(file_id)
391
1122
return self._byid[file_id]
392
1123
except KeyError:
394
raise BzrError("can't look up file_id None")
396
raise BzrError("file_id {%s} not in inventory" % file_id)
1124
# really we're passing an inventory, not a tree...
1125
raise errors.NoSuchId(self, file_id)
399
1127
def get_file_kind(self, file_id):
1128
file_id = osutils.safe_file_id(file_id)
400
1129
return self._byid[file_id].kind
402
1131
def get_child(self, parent_id, filename):
1132
parent_id = osutils.safe_file_id(parent_id)
403
1133
return self[parent_id].children.get(filename)
1135
def _add_child(self, entry):
1136
"""Add an entry to the inventory, without adding it to its parent"""
1137
if entry.file_id in self._byid:
1138
raise BzrError("inventory already contains entry with id {%s}" %
1140
self._byid[entry.file_id] = entry
1141
for child in getattr(entry, 'children', {}).itervalues():
1142
self._add_child(child)
406
1145
def add(self, entry):
407
1146
"""Add entry to inventory.
409
1148
To add a file to a branch ready to be committed, use Branch.add,
1151
Returns the new entry object.
411
1153
if entry.file_id in self._byid:
412
raise BzrError("inventory already contains entry with id {%s}" % entry.file_id)
414
if entry.parent_id == ROOT_ID or entry.parent_id is None:
415
entry.parent_id = self.root.file_id
418
parent = self._byid[entry.parent_id]
420
raise BzrError("parent_id {%s} not in inventory" % entry.parent_id)
422
if parent.children.has_key(entry.name):
423
raise BzrError("%s is already versioned" %
424
appendpath(self.id2path(parent.file_id), entry.name))
426
self._byid[entry.file_id] = entry
427
parent.children[entry.name] = entry
430
def add_path(self, relpath, kind, file_id=None):
1154
raise errors.DuplicateFileId(entry.file_id,
1155
self._byid[entry.file_id])
1157
if entry.parent_id is None:
1158
assert self.root is None and len(self._byid) == 0
1162
parent = self._byid[entry.parent_id]
1164
raise BzrError("parent_id {%s} not in inventory" %
1167
if entry.name in parent.children:
1168
raise BzrError("%s is already versioned" %
1169
osutils.pathjoin(self.id2path(parent.file_id),
1170
entry.name).encode('utf-8'))
1171
parent.children[entry.name] = entry
1172
return self._add_child(entry)
1174
def add_path(self, relpath, kind, file_id=None, parent_id=None):
431
1175
"""Add entry from a path.
433
The immediate parent must already be versioned"""
434
from bzrlib.branch import gen_file_id
1177
The immediate parent must already be versioned.
1179
Returns the new entry object."""
436
parts = bzrlib.osutils.splitpath(relpath)
1181
parts = osutils.splitpath(relpath)
437
1183
if len(parts) == 0:
438
raise BzrError("cannot re-add root of inventory")
441
file_id = gen_file_id(relpath)
443
parent_path = parts[:-1]
444
parent_id = self.path2id(parent_path)
445
if parent_id == None:
446
raise NotVersionedError(parent_path)
448
ie = InventoryEntry(file_id, parts[-1],
449
kind=kind, parent_id=parent_id)
1185
file_id = generate_ids.gen_root_id()
1187
file_id = osutils.safe_file_id(file_id)
1188
self.root = InventoryDirectory(file_id, '', None)
1189
self._byid = {self.root.file_id: self.root}
1192
parent_path = parts[:-1]
1193
parent_id = self.path2id(parent_path)
1194
if parent_id is None:
1195
raise errors.NotVersionedError(path=parent_path)
1196
ie = make_entry(kind, parts[-1], parent_id, file_id)
450
1197
return self.add(ie)
453
1199
def __delitem__(self, file_id):
454
1200
"""Remove entry by id.
456
1202
>>> inv = Inventory()
457
>>> inv.add(InventoryEntry('123', 'foo.c', 'file', ROOT_ID))
1203
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1204
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
458
1205
>>> '123' in inv
460
1207
>>> del inv['123']
461
1208
>>> '123' in inv
1211
file_id = osutils.safe_file_id(file_id)
464
1212
ie = self[file_id]
466
assert self[ie.parent_id].children[ie.name] == ie
1214
assert ie.parent_id is None or \
1215
self[ie.parent_id].children[ie.name] == ie
468
# TODO: Test deleting all children; maybe hoist to a separate
470
if ie.kind == 'directory':
471
for cie in ie.children.values():
472
del self[cie.file_id]
475
1217
del self._byid[file_id]
476
del self[ie.parent_id].children[ie.name]
479
def to_element(self):
480
"""Convert to XML Element"""
481
from bzrlib.xml import Element
483
e = Element('inventory')
485
if self.root.file_id not in (None, ROOT_ID):
486
e.set('file_id', self.root.file_id)
487
for path, ie in self.iter_entries():
488
e.append(ie.to_element())
492
def from_element(cls, elt):
493
"""Construct from XML Element
495
>>> inv = Inventory()
496
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c', 'file', ROOT_ID))
497
>>> elt = inv.to_element()
498
>>> inv2 = Inventory.from_element(elt)
502
# XXXX: doctest doesn't run this properly under python2.3
503
assert elt.tag == 'inventory'
504
root_id = elt.get('file_id') or ROOT_ID
507
ie = InventoryEntry.from_element(e)
508
if ie.parent_id == ROOT_ID:
509
ie.parent_id = root_id
513
from_element = classmethod(from_element)
1218
if ie.parent_id is not None:
1219
del self[ie.parent_id].children[ie.name]
516
1221
def __eq__(self, other):
517
1222
"""Compare two sets by comparing their contents.
added
removed
bzrlib/inventory.py
