48
36
An InventoryEntry has the following fields, which are also
49
37
present in the XML inventory-entry element:
54
(within the parent directory)
57
file_id of the parent directory, or ROOT_ID
60
the revision_id in which this variation of this file was
64
Indicates that this file should be executable on systems
68
sha-1 of the text of the file
71
size in bytes of the text of the file
73
(reading a version 4 tree created a text_id field.)
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.
75
54
>>> i = Inventory()
78
>>> i.add(InventoryDirectory('123', 'src', ROOT_ID))
79
InventoryDirectory('123', 'src', parent_id='TREE_ROOT')
80
>>> i.add(InventoryFile('2323', 'hello.c', parent_id='123'))
81
InventoryFile('2323', 'hello.c', parent_id='123')
82
>>> shouldbe = {0: 'src', 1: pathjoin('src','hello.c')}
83
>>> for ix, j in enumerate(i.iter_entries()):
84
... print (j[0] == shouldbe[ix], j[1])
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():
86
(True, InventoryDirectory('123', 'src', parent_id='TREE_ROOT'))
87
(True, InventoryFile('2323', 'hello.c', parent_id='123'))
88
>>> i.add(InventoryFile('2323', 'bye.c', '123'))
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'))
89
65
Traceback (most recent call last):
91
67
BzrError: inventory already contains entry with id {2323}
92
>>> i.add(InventoryFile('2324', 'bye.c', '123'))
93
InventoryFile('2324', 'bye.c', parent_id='123')
94
>>> i.add(InventoryDirectory('2325', 'wibble', '123'))
95
InventoryDirectory('2325', 'wibble', parent_id='123')
68
>>> i.add(InventoryEntry('2324', 'bye.c', 'file', '123'))
69
>>> i.add(InventoryEntry('2325', 'wibble', 'directory', '123'))
96
70
>>> i.path2id('src/wibble')
100
>>> i.add(InventoryFile('2326', 'wibble.c', '2325'))
101
InventoryFile('2326', 'wibble.c', parent_id='2325')
74
>>> i.add(InventoryEntry('2326', 'wibble.c', 'file', '2325'))
103
InventoryFile('2326', 'wibble.c', parent_id='2325')
104
>>> for path, entry in i.iter_entries():
106
... assert i.path2id(path)
76
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
77
>>> for j in i.iter_entries():
79
... assert i.path2id(j[0])
112
85
src/wibble/wibble.c
113
86
>>> i.id2path('2326')
114
87
'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.
117
97
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
118
'text_id', 'parent_id', 'children', 'executable',
121
def _add_text_to_weave(self, new_lines, parents, weave_store, transaction):
122
versionedfile = weave_store.get_weave_or_empty(self.file_id,
124
versionedfile.add_lines(self.revision, parents, new_lines)
126
def detect_changes(self, old_entry):
127
"""Return a (text_modified, meta_modified) from this to old_entry.
129
_read_tree_state must have been called on self and old_entry prior to
130
calling detect_changes.
134
def diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
135
output_to, reverse=False):
136
"""Perform a diff from this to to_entry.
138
text_diff will be used for textual difference calculation.
139
This is a template method, override _diff in child classes.
141
self._read_tree_state(tree.id2path(self.file_id), tree)
143
# cannot diff from one kind to another - you must do a removal
144
# and an addif they do not match.
145
assert self.kind == to_entry.kind
146
to_entry._read_tree_state(to_tree.id2path(to_entry.file_id),
148
self._diff(text_diff, from_label, tree, to_label, to_entry, to_tree,
151
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
152
output_to, reverse=False):
153
"""Perform a diff between two entries of the same kind."""
155
def find_previous_heads(self, previous_inventories,
156
versioned_file_store,
159
"""Return the revisions and entries that directly preceed this.
161
Returned as a map from revision to inventory entry.
163
This is a map containing the file revisions in all parents
164
for which the file exists, and its revision is not a parent of
165
any other. If the file is new, the set will be empty.
167
:param versioned_file_store: A store where ancestry data on this
168
file id can be queried.
169
:param transaction: The transaction that queries to the versioned
170
file store should be completed under.
171
:param entry_vf: The entry versioned file, if its already available.
173
def get_ancestors(weave, entry):
174
return set(weave.get_ancestry(entry.revision))
175
# revision:ie mapping for each ie found in previous_inventories.
177
# revision:ie mapping with one revision for each head.
179
# revision: ancestor list for each head
181
# identify candidate head revision ids.
182
for inv in previous_inventories:
183
if self.file_id in inv:
184
ie = inv[self.file_id]
185
assert ie.file_id == self.file_id
186
if ie.revision in candidates:
187
# same revision value in two different inventories:
188
# correct possible inconsistencies:
189
# * there was a bug in revision updates with 'x' bit
192
if candidates[ie.revision].executable != ie.executable:
193
candidates[ie.revision].executable = False
194
ie.executable = False
195
except AttributeError:
197
# must now be the same.
198
assert candidates[ie.revision] == ie
200
# add this revision as a candidate.
201
candidates[ie.revision] = ie
203
# common case optimisation
204
if len(candidates) == 1:
205
# if there is only one candidate revision found
206
# then we can opening the versioned file to access ancestry:
207
# there cannot be any ancestors to eliminate when there is
208
# only one revision available.
209
heads[ie.revision] = ie
212
# eliminate ancestors amongst the available candidates:
213
# heads are those that are not an ancestor of any other candidate
214
# - this provides convergence at a per-file level.
215
for ie in candidates.values():
216
# may be an ancestor of a known head:
217
already_present = 0 != len(
218
[head for head in heads
219
if ie.revision in head_ancestors[head]])
221
# an ancestor of an analyzed candidate.
223
# not an ancestor of a known head:
224
# load the versioned file for this file id if needed
226
entry_vf = versioned_file_store.get_weave_or_empty(
227
self.file_id, transaction)
228
ancestors = get_ancestors(entry_vf, ie)
229
# may knock something else out:
230
check_heads = list(heads.keys())
231
for head in check_heads:
232
if head in ancestors:
233
# this previously discovered 'head' is not
234
# really a head - its an ancestor of the newly
237
head_ancestors[ie.revision] = ancestors
238
heads[ie.revision] = ie
241
def get_tar_item(self, root, dp, now, tree):
242
"""Get a tarfile item and a file stream for its content."""
243
item = tarfile.TarInfo(pathjoin(root, dp))
244
# TODO: would be cool to actually set it to the timestamp of the
245
# revision it was last changed
247
fileobj = self._put_in_tar(item, tree)
251
"""Return true if the object this entry represents has textual data.
253
Note that textual data includes binary content.
255
Also note that all entries get weave files created for them.
256
This attribute is primarily used when upgrading from old trees that
257
did not have the weave index for all inventory entries.
261
def __init__(self, file_id, name, parent_id, text_id=None):
98
'text_id', 'parent_id', 'children', ]
100
def __init__(self, file_id, name, kind, parent_id, text_id=None):
262
101
"""Create an InventoryEntry
264
103
The filename must be a single component, relative to the
265
104
parent directory; it cannot be a whole path or relative name.
267
>>> e = InventoryFile('123', 'hello.c', ROOT_ID)
106
>>> e = InventoryEntry('123', 'hello.c', 'file', ROOT_ID)
272
>>> e = InventoryFile('123', 'src/hello.c', ROOT_ID)
111
>>> e = InventoryEntry('123', 'src/hello.c', 'file', ROOT_ID)
273
112
Traceback (most recent call last):
274
InvalidEntryName: Invalid entry name: src/hello.c
113
BzrCheckError: InventoryEntry name 'src/hello.c' is invalid
276
assert isinstance(name, basestring), name
277
115
if '/' in name or '\\' in name:
278
raise InvalidEntryName(name=name)
279
self.executable = False
116
raise BzrCheckError('InventoryEntry name %r is invalid' % name)
281
118
self.text_sha1 = None
282
119
self.text_size = None
283
121
self.file_id = file_id
285
124
self.text_id = text_id
286
125
self.parent_id = parent_id
287
self.symlink_target = None
289
def kind_character(self):
290
"""Return a short kind indicator useful for appending to names."""
291
raise BzrError('unknown kind %r' % self.kind)
293
known_kinds = ('file', 'directory', 'symlink', 'root_directory')
295
def _put_in_tar(self, item, tree):
296
"""populate item for stashing in a tar, and return the content stream.
298
If no content is available, return None.
300
raise BzrError("don't know how to export {%s} of kind %r" %
301
(self.file_id, self.kind))
303
def put_on_disk(self, dest, dp, tree):
304
"""Create a representation of self on disk in the prefix dest.
306
This is a template method - implement _put_on_disk in subclasses.
308
fullpath = pathjoin(dest, dp)
309
self._put_on_disk(fullpath, tree)
310
mutter(" export {%s} kind %s to %s", self.file_id,
313
def _put_on_disk(self, fullpath, tree):
314
"""Put this entry onto disk at fullpath, from tree tree."""
315
raise BzrError("don't know how to export {%s} of kind %r" % (self.file_id, self.kind))
126
if kind == 'directory':
131
raise BzrError("unhandled entry kind %r" % kind)
317
135
def sorted_children(self):
318
136
l = self.children.items()
323
def versionable_kind(kind):
324
return kind in ('file', 'directory', 'symlink')
326
def check(self, checker, rev_id, inv, tree):
327
"""Check this inventory entry is intact.
329
This is a template method, override _check for kind specific
332
:param checker: Check object providing context for the checks;
333
can be used to find out what parts of the repository have already
335
:param rev_id: Revision id from which this InventoryEntry was loaded.
336
Not necessarily the last-changed revision for this file.
337
:param inv: Inventory from which the entry was loaded.
338
:param tree: RevisionTree for this entry.
340
if self.parent_id != None:
341
if not inv.has_id(self.parent_id):
342
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
343
% (self.parent_id, rev_id))
344
self._check(checker, rev_id, tree)
346
def _check(self, checker, rev_id, tree):
347
"""Check this inventory entry for kind specific errors."""
348
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
353
"""Clone this inventory entry."""
354
raise NotImplementedError
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
356
def _get_snapshot_change(self, previous_entries):
357
if len(previous_entries) > 1:
359
elif len(previous_entries) == 0:
362
return 'modified/renamed/reparented'
364
151
def __repr__(self):
365
return ("%s(%r, %r, parent_id=%r)"
152
return ("%s(%r, %r, kind=%r, parent_id=%r)"
366
153
% (self.__class__.__name__,
371
def snapshot(self, revision, path, previous_entries,
372
work_tree, weave_store, transaction):
373
"""Make a snapshot of this entry which may or may not have changed.
375
This means that all its fields are populated, that it has its
376
text stored in the text store or weave.
378
mutter('new parents of %s are %r', path, previous_entries)
379
self._read_tree_state(path, work_tree)
380
if len(previous_entries) == 1:
381
# cannot be unchanged unless there is only one parent file rev.
382
parent_ie = previous_entries.values()[0]
383
if self._unchanged(parent_ie):
384
mutter("found unchanged entry")
385
self.revision = parent_ie.revision
387
return self.snapshot_revision(revision, previous_entries,
388
work_tree, weave_store, transaction)
390
def snapshot_revision(self, revision, previous_entries, work_tree,
391
weave_store, transaction):
392
"""Record this revision unconditionally."""
393
mutter('new revision for {%s}', self.file_id)
394
self.revision = revision
395
change = self._get_snapshot_change(previous_entries)
396
self._snapshot_text(previous_entries, work_tree, weave_store,
400
def _snapshot_text(self, file_parents, work_tree, weave_store, transaction):
401
"""Record the 'text' of this entry, whatever form that takes.
403
This default implementation simply adds an empty text.
405
mutter('storing file {%s} in revision {%s}',
406
self.file_id, self.revision)
407
self._add_text_to_weave([], file_parents.keys(), weave_store, transaction)
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)
409
214
def __eq__(self, other):
410
215
if not isinstance(other, InventoryEntry):
411
216
return NotImplemented
413
return ((self.file_id == other.file_id)
414
and (self.name == other.name)
415
and (other.symlink_target == self.symlink_target)
416
and (self.text_sha1 == other.text_sha1)
417
and (self.text_size == other.text_size)
418
and (self.text_id == other.text_id)
419
and (self.parent_id == other.parent_id)
420
and (self.kind == other.kind)
421
and (self.revision == other.revision)
422
and (self.executable == other.executable)
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)
425
227
def __ne__(self, other):
426
228
return not (self == other)
477
248
and (self.children == other.children)
480
class InventoryDirectory(InventoryEntry):
481
"""A directory in an inventory."""
483
def _check(self, checker, rev_id, tree):
484
"""See InventoryEntry._check"""
485
if self.text_sha1 != None or self.text_size != None or self.text_id != None:
486
raise BzrCheckError('directory {%s} has text in revision {%s}'
487
% (self.file_id, rev_id))
490
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
491
other.revision = self.revision
492
# note that children are *not* copied; they're pulled across when
496
def __init__(self, file_id, name, parent_id):
497
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
499
self.kind = 'directory'
501
def kind_character(self):
502
"""See InventoryEntry.kind_character."""
505
def _put_in_tar(self, item, tree):
506
"""See InventoryEntry._put_in_tar."""
507
item.type = tarfile.DIRTYPE
514
def _put_on_disk(self, fullpath, tree):
515
"""See InventoryEntry._put_on_disk."""
519
class InventoryFile(InventoryEntry):
520
"""A file in an inventory."""
522
def _check(self, checker, tree_revision_id, tree):
523
"""See InventoryEntry._check"""
524
t = (self.file_id, self.revision)
525
if t in checker.checked_texts:
526
prev_sha = checker.checked_texts[t]
527
if prev_sha != self.text_sha1:
528
raise BzrCheckError('mismatched sha1 on {%s} in {%s}' %
529
(self.file_id, tree_revision_id))
531
checker.repeated_text_cnt += 1
534
if self.file_id not in checker.checked_weaves:
535
mutter('check weave {%s}', self.file_id)
536
w = tree.get_weave(self.file_id)
537
# Not passing a progress bar, because it creates a new
538
# progress, which overwrites the current progress,
539
# and doesn't look nice
541
checker.checked_weaves[self.file_id] = True
543
w = tree.get_weave(self.file_id)
545
mutter('check version {%s} of {%s}', tree_revision_id, self.file_id)
546
checker.checked_text_cnt += 1
547
# We can't check the length, because Weave doesn't store that
548
# information, and the whole point of looking at the weave's
549
# sha1sum is that we don't have to extract the text.
550
if self.text_sha1 != w.get_sha1(self.revision):
551
raise BzrCheckError('text {%s} version {%s} wrong sha1'
552
% (self.file_id, self.revision))
553
checker.checked_texts[t] = self.text_sha1
556
other = InventoryFile(self.file_id, self.name, self.parent_id)
557
other.executable = self.executable
558
other.text_id = self.text_id
559
other.text_sha1 = self.text_sha1
560
other.text_size = self.text_size
561
other.revision = self.revision
564
def detect_changes(self, old_entry):
565
"""See InventoryEntry.detect_changes."""
566
assert self.text_sha1 != None
567
assert old_entry.text_sha1 != None
568
text_modified = (self.text_sha1 != old_entry.text_sha1)
569
meta_modified = (self.executable != old_entry.executable)
570
return text_modified, meta_modified
572
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
573
output_to, reverse=False):
574
"""See InventoryEntry._diff."""
575
from_text = tree.get_file(self.file_id).readlines()
577
to_text = to_tree.get_file(to_entry.file_id).readlines()
581
text_diff(from_label, from_text,
582
to_label, to_text, output_to)
584
text_diff(to_label, to_text,
585
from_label, from_text, output_to)
588
"""See InventoryEntry.has_text."""
591
def __init__(self, file_id, name, parent_id):
592
super(InventoryFile, self).__init__(file_id, name, parent_id)
595
def kind_character(self):
596
"""See InventoryEntry.kind_character."""
599
def _put_in_tar(self, item, tree):
600
"""See InventoryEntry._put_in_tar."""
601
item.type = tarfile.REGTYPE
602
fileobj = tree.get_file(self.file_id)
603
item.size = self.text_size
604
if tree.is_executable(self.file_id):
610
def _put_on_disk(self, fullpath, tree):
611
"""See InventoryEntry._put_on_disk."""
612
pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
613
if tree.is_executable(self.file_id):
614
os.chmod(fullpath, 0755)
616
def _read_tree_state(self, path, work_tree):
617
"""See InventoryEntry._read_tree_state."""
618
self.text_sha1 = work_tree.get_file_sha1(self.file_id)
619
self.executable = work_tree.is_executable(self.file_id)
621
def _forget_tree_state(self):
622
self.text_sha1 = None
623
self.executable = None
625
def _snapshot_text(self, file_parents, work_tree, weave_store, transaction):
626
"""See InventoryEntry._snapshot_text."""
627
mutter('storing file {%s} in revision {%s}',
628
self.file_id, self.revision)
629
# special case to avoid diffing on renames or
631
if (len(file_parents) == 1
632
and self.text_sha1 == file_parents.values()[0].text_sha1
633
and self.text_size == file_parents.values()[0].text_size):
634
previous_ie = file_parents.values()[0]
635
versionedfile = weave_store.get_weave(self.file_id, transaction)
636
versionedfile.clone_text(self.revision, previous_ie.revision, file_parents.keys())
638
new_lines = work_tree.get_file(self.file_id).readlines()
639
self._add_text_to_weave(new_lines, file_parents.keys(), weave_store,
641
self.text_sha1 = sha_strings(new_lines)
642
self.text_size = sum(map(len, new_lines))
645
def _unchanged(self, previous_ie):
646
"""See InventoryEntry._unchanged."""
647
compatible = super(InventoryFile, self)._unchanged(previous_ie)
648
if self.text_sha1 != previous_ie.text_sha1:
651
# FIXME: 20050930 probe for the text size when getting sha1
652
# in _read_tree_state
653
self.text_size = previous_ie.text_size
654
if self.executable != previous_ie.executable:
659
class InventoryLink(InventoryEntry):
660
"""A file in an inventory."""
662
__slots__ = ['symlink_target']
664
def _check(self, checker, rev_id, tree):
665
"""See InventoryEntry._check"""
666
if self.text_sha1 != None or self.text_size != None or self.text_id != None:
667
raise BzrCheckError('symlink {%s} has text in revision {%s}'
668
% (self.file_id, rev_id))
669
if self.symlink_target == None:
670
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
671
% (self.file_id, rev_id))
674
other = InventoryLink(self.file_id, self.name, self.parent_id)
675
other.symlink_target = self.symlink_target
676
other.revision = self.revision
679
def detect_changes(self, old_entry):
680
"""See InventoryEntry.detect_changes."""
681
# FIXME: which _modified field should we use ? RBC 20051003
682
text_modified = (self.symlink_target != old_entry.symlink_target)
684
mutter(" symlink target changed")
685
meta_modified = False
686
return text_modified, meta_modified
688
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
689
output_to, reverse=False):
690
"""See InventoryEntry._diff."""
691
from_text = self.symlink_target
692
if to_entry is not None:
693
to_text = to_entry.symlink_target
698
print >>output_to, '=== target changed %r => %r' % (from_text, to_text)
701
print >>output_to, '=== target was %r' % self.symlink_target
703
print >>output_to, '=== target is %r' % self.symlink_target
705
def __init__(self, file_id, name, parent_id):
706
super(InventoryLink, self).__init__(file_id, name, parent_id)
707
self.kind = 'symlink'
709
def kind_character(self):
710
"""See InventoryEntry.kind_character."""
713
def _put_in_tar(self, item, tree):
714
"""See InventoryEntry._put_in_tar."""
715
item.type = tarfile.SYMTYPE
719
item.linkname = self.symlink_target
722
def _put_on_disk(self, fullpath, tree):
723
"""See InventoryEntry._put_on_disk."""
725
os.symlink(self.symlink_target, fullpath)
727
raise BzrError("Failed to create symlink %r -> %r, error: %s" % (fullpath, self.symlink_target, e))
729
def _read_tree_state(self, path, work_tree):
730
"""See InventoryEntry._read_tree_state."""
731
self.symlink_target = work_tree.get_symlink_target(self.file_id)
733
def _forget_tree_state(self):
734
self.symlink_target = None
736
def _unchanged(self, previous_ie):
737
"""See InventoryEntry._unchanged."""
738
compatible = super(InventoryLink, self)._unchanged(previous_ie)
739
if self.symlink_target != previous_ie.symlink_target:
744
252
class Inventory(object):
745
253
"""Inventory of versioned files in a tree.
added
removed
bzrlib/inventory.py
