1
# Copyright (C) 2005, 2006 Canonical Ltd
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
# FIXME: This refactoring of the workingtree code doesn't seem to keep
18
# the WorkingTree's copy of the inventory in sync with the branch. The
19
# branch modifies its working inventory when it does a commit to make
20
# missing files permanently removed.
22
# TODO: Maybe also keep the full path of the entry, and the children?
23
# But those depend on its position within a particular inventory, and
24
# it would be nice not to need to hold the backpointer here.
26
# This should really be an id randomly assigned when the tree is
27
# created, but it's not for now.
39
from bzrlib.osutils import (pumpfile, quotefn, splitpath, joinpath,
40
pathjoin, sha_strings)
41
from bzrlib.errors import (NotVersionedError, InvalidEntryName,
42
BzrError, BzrCheckError, BinaryFile)
43
from bzrlib.trace import mutter
46
class InventoryEntry(object):
47
"""Description of a versioned file.
49
An InventoryEntry has the following fields, which are also
50
present in the XML inventory-entry element:
55
(within the parent directory)
58
file_id of the parent directory, or ROOT_ID
61
the revision_id in which this variation of this file was
65
Indicates that this file should be executable on systems
69
sha-1 of the text of the file
72
size in bytes of the text of the file
74
(reading a version 4 tree created a text_id field.)
79
>>> i.add(InventoryDirectory('123', 'src', ROOT_ID))
80
InventoryDirectory('123', 'src', parent_id='TREE_ROOT')
81
>>> i.add(InventoryFile('2323', 'hello.c', parent_id='123'))
82
InventoryFile('2323', 'hello.c', parent_id='123')
83
>>> shouldbe = {0: 'src', 1: pathjoin('src','hello.c')}
84
>>> for ix, j in enumerate(i.iter_entries()):
85
... print (j[0] == shouldbe[ix], j[1])
87
(True, InventoryDirectory('123', 'src', parent_id='TREE_ROOT'))
88
(True, InventoryFile('2323', 'hello.c', parent_id='123'))
89
>>> i.add(InventoryFile('2323', 'bye.c', '123'))
90
Traceback (most recent call last):
92
BzrError: inventory already contains entry with id {2323}
93
>>> i.add(InventoryFile('2324', 'bye.c', '123'))
94
InventoryFile('2324', 'bye.c', parent_id='123')
95
>>> i.add(InventoryDirectory('2325', 'wibble', '123'))
96
InventoryDirectory('2325', 'wibble', parent_id='123')
97
>>> i.path2id('src/wibble')
101
>>> i.add(InventoryFile('2326', 'wibble.c', '2325'))
102
InventoryFile('2326', 'wibble.c', parent_id='2325')
104
InventoryFile('2326', 'wibble.c', parent_id='2325')
105
>>> for path, entry in i.iter_entries():
107
... assert i.path2id(path)
114
>>> i.id2path('2326')
115
'src/wibble/wibble.c'
118
# Constants returned by describe_change()
120
# TODO: These should probably move to some kind of FileChangeDescription
121
# class; that's like what's inside a TreeDelta but we want to be able to
122
# generate them just for one file at a time.
124
MODIFIED_AND_RENAMED = 'modified and renamed'
126
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
127
'text_id', 'parent_id', 'children', 'executable',
130
def _add_text_to_weave(self, new_lines, parents, weave_store, transaction):
131
versionedfile = weave_store.get_weave_or_empty(self.file_id,
133
versionedfile.add_lines(self.revision, parents, new_lines)
134
versionedfile.clear_cache()
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 preceed 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))
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.revision in candidates:
197
# same revision value in two different inventories:
198
# correct possible inconsistencies:
199
# * there was a bug in revision updates with 'x' bit
202
if candidates[ie.revision].executable != ie.executable:
203
candidates[ie.revision].executable = False
204
ie.executable = False
205
except AttributeError:
207
# must now be the same.
208
assert candidates[ie.revision] == ie
210
# add this revision as a candidate.
211
candidates[ie.revision] = ie
213
# common case optimisation
214
if len(candidates) == 1:
215
# if there is only one candidate revision found
216
# then we can opening the versioned file to access ancestry:
217
# there cannot be any ancestors to eliminate when there is
218
# only one revision available.
219
heads[ie.revision] = ie
222
# eliminate ancestors amongst the available candidates:
223
# heads are those that are not an ancestor of any other candidate
224
# - this provides convergence at a per-file level.
225
for ie in candidates.values():
226
# may be an ancestor of a known head:
227
already_present = 0 != len(
228
[head for head in heads
229
if ie.revision in head_ancestors[head]])
231
# an ancestor of an analyzed candidate.
233
# not an ancestor of a known head:
234
# load the versioned file for this file id if needed
236
entry_vf = versioned_file_store.get_weave_or_empty(
237
self.file_id, transaction)
238
ancestors = get_ancestors(entry_vf, ie)
239
# may knock something else out:
240
check_heads = list(heads.keys())
241
for head in check_heads:
242
if head in ancestors:
243
# this previously discovered 'head' is not
244
# really a head - its an ancestor of the newly
247
head_ancestors[ie.revision] = ancestors
248
heads[ie.revision] = ie
251
def get_tar_item(self, root, dp, now, tree):
252
"""Get a tarfile item and a file stream for its content."""
253
item = tarfile.TarInfo(pathjoin(root, dp))
254
# TODO: would be cool to actually set it to the timestamp of the
255
# revision it was last changed
257
fileobj = self._put_in_tar(item, tree)
261
"""Return true if the object this entry represents has textual data.
263
Note that textual data includes binary content.
265
Also note that all entries get weave files created for them.
266
This attribute is primarily used when upgrading from old trees that
267
did not have the weave index for all inventory entries.
271
def __init__(self, file_id, name, parent_id, text_id=None):
272
"""Create an InventoryEntry
274
The filename must be a single component, relative to the
275
parent directory; it cannot be a whole path or relative name.
277
>>> e = InventoryFile('123', 'hello.c', ROOT_ID)
282
>>> e = InventoryFile('123', 'src/hello.c', ROOT_ID)
283
Traceback (most recent call last):
284
InvalidEntryName: Invalid entry name: src/hello.c
286
assert isinstance(name, basestring), name
287
if '/' in name or '\\' in name:
288
raise InvalidEntryName(name=name)
289
self.executable = False
291
self.text_sha1 = None
292
self.text_size = None
293
self.file_id = file_id
295
self.text_id = text_id
296
self.parent_id = parent_id
297
self.symlink_target = None
299
def kind_character(self):
300
"""Return a short kind indicator useful for appending to names."""
301
raise BzrError('unknown kind %r' % self.kind)
303
known_kinds = ('file', 'directory', 'symlink')
305
def _put_in_tar(self, item, tree):
306
"""populate item for stashing in a tar, and return the content stream.
308
If no content is available, return None.
310
raise BzrError("don't know how to export {%s} of kind %r" %
311
(self.file_id, self.kind))
313
def put_on_disk(self, dest, dp, tree):
314
"""Create a representation of self on disk in the prefix dest.
316
This is a template method - implement _put_on_disk in subclasses.
318
fullpath = pathjoin(dest, dp)
319
self._put_on_disk(fullpath, tree)
320
mutter(" export {%s} kind %s to %s", self.file_id,
323
def _put_on_disk(self, fullpath, tree):
324
"""Put this entry onto disk at fullpath, from tree tree."""
325
raise BzrError("don't know how to export {%s} of kind %r" % (self.file_id, self.kind))
327
def sorted_children(self):
328
l = self.children.items()
333
def versionable_kind(kind):
334
return kind in ('file', 'directory', 'symlink')
336
def check(self, checker, rev_id, inv, tree):
337
"""Check this inventory entry is intact.
339
This is a template method, override _check for kind specific
342
:param checker: Check object providing context for the checks;
343
can be used to find out what parts of the repository have already
345
:param rev_id: Revision id from which this InventoryEntry was loaded.
346
Not necessarily the last-changed revision for this file.
347
:param inv: Inventory from which the entry was loaded.
348
:param tree: RevisionTree for this entry.
350
if self.parent_id != None:
351
if not inv.has_id(self.parent_id):
352
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
353
% (self.parent_id, rev_id))
354
self._check(checker, rev_id, tree)
356
def _check(self, checker, rev_id, tree):
357
"""Check this inventory entry for kind specific errors."""
358
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
362
"""Clone this inventory entry."""
363
raise NotImplementedError
366
def describe_change(old_entry, new_entry):
367
"""Describe the change between old_entry and this.
369
This smells of being an InterInventoryEntry situation, but as its
370
the first one, we're making it a static method for now.
372
An entry with a different parent, or different name is considered
373
to be renamed. Reparenting is an internal detail.
374
Note that renaming the parent does not trigger a rename for the
377
# TODO: Perhaps return an object rather than just a string
378
if old_entry is new_entry:
379
# also the case of both being None
381
elif old_entry is None:
383
elif new_entry is None:
385
text_modified, meta_modified = new_entry.detect_changes(old_entry)
386
if text_modified or meta_modified:
390
# TODO 20060511 (mbp, rbc) factor out 'detect_rename' here.
391
if old_entry.parent_id != new_entry.parent_id:
393
elif old_entry.name != new_entry.name:
397
if renamed and not modified:
398
return InventoryEntry.RENAMED
399
if modified and not renamed:
401
if modified and renamed:
402
return InventoryEntry.MODIFIED_AND_RENAMED
406
return ("%s(%r, %r, parent_id=%r)"
407
% (self.__class__.__name__,
412
def snapshot(self, revision, path, previous_entries,
413
work_tree, weave_store, transaction):
414
"""Make a snapshot of this entry which may or may not have changed.
416
This means that all its fields are populated, that it has its
417
text stored in the text store or weave.
419
mutter('new parents of %s are %r', path, previous_entries)
420
self._read_tree_state(path, work_tree)
421
if len(previous_entries) == 1:
422
# cannot be unchanged unless there is only one parent file rev.
423
parent_ie = previous_entries.values()[0]
424
if self._unchanged(parent_ie):
425
mutter("found unchanged entry")
426
self.revision = parent_ie.revision
428
return self._snapshot_into_revision(revision, previous_entries,
429
work_tree, weave_store, transaction)
431
def _snapshot_into_revision(self, revision, previous_entries, work_tree,
432
weave_store, transaction):
433
"""Record this revision unconditionally into a store.
435
The entry's last-changed revision property (`revision`) is updated to
436
that of the new revision.
438
:param revision: id of the new revision that is being recorded.
440
:returns: String description of the commit (e.g. "merged", "modified"), etc.
442
mutter('new revision {%s} for {%s}', revision, self.file_id)
443
self.revision = revision
444
self._snapshot_text(previous_entries, work_tree, weave_store,
447
def _snapshot_text(self, file_parents, work_tree, weave_store, transaction):
448
"""Record the 'text' of this entry, whatever form that takes.
450
This default implementation simply adds an empty text.
452
mutter('storing file {%s} in revision {%s}',
453
self.file_id, self.revision)
454
self._add_text_to_weave([], file_parents.keys(), weave_store, transaction)
456
def __eq__(self, other):
457
if not isinstance(other, InventoryEntry):
458
return NotImplemented
460
return ((self.file_id == other.file_id)
461
and (self.name == other.name)
462
and (other.symlink_target == self.symlink_target)
463
and (self.text_sha1 == other.text_sha1)
464
and (self.text_size == other.text_size)
465
and (self.text_id == other.text_id)
466
and (self.parent_id == other.parent_id)
467
and (self.kind == other.kind)
468
and (self.revision == other.revision)
469
and (self.executable == other.executable)
472
def __ne__(self, other):
473
return not (self == other)
476
raise ValueError('not hashable')
478
def _unchanged(self, previous_ie):
479
"""Has this entry changed relative to previous_ie.
481
This method should be overriden in child classes.
484
# different inv parent
485
if previous_ie.parent_id != self.parent_id:
488
elif previous_ie.name != self.name:
492
def _read_tree_state(self, path, work_tree):
493
"""Populate fields in the inventory entry from the given tree.
495
Note that this should be modified to be a noop on virtual trees
496
as all entries created there are prepopulated.
498
# TODO: Rather than running this manually, we should check the
499
# working sha1 and other expensive properties when they're
500
# first requested, or preload them if they're already known
501
pass # nothing to do by default
503
def _forget_tree_state(self):
507
class InventoryDirectory(InventoryEntry):
508
"""A directory in an inventory."""
510
def _check(self, checker, rev_id, tree):
511
"""See InventoryEntry._check"""
512
if self.text_sha1 != None or self.text_size != None or self.text_id != None:
513
raise BzrCheckError('directory {%s} has text in revision {%s}'
514
% (self.file_id, rev_id))
517
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
518
other.revision = self.revision
519
# note that children are *not* copied; they're pulled across when
523
def __init__(self, file_id, name, parent_id):
524
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
526
self.kind = 'directory'
528
def kind_character(self):
529
"""See InventoryEntry.kind_character."""
532
def _put_in_tar(self, item, tree):
533
"""See InventoryEntry._put_in_tar."""
534
item.type = tarfile.DIRTYPE
541
def _put_on_disk(self, fullpath, tree):
542
"""See InventoryEntry._put_on_disk."""
546
class InventoryFile(InventoryEntry):
547
"""A file in an inventory."""
549
def _check(self, checker, tree_revision_id, tree):
550
"""See InventoryEntry._check"""
551
t = (self.file_id, self.revision)
552
if t in checker.checked_texts:
553
prev_sha = checker.checked_texts[t]
554
if prev_sha != self.text_sha1:
555
raise BzrCheckError('mismatched sha1 on {%s} in {%s}' %
556
(self.file_id, tree_revision_id))
558
checker.repeated_text_cnt += 1
561
if self.file_id not in checker.checked_weaves:
562
mutter('check weave {%s}', self.file_id)
563
w = tree.get_weave(self.file_id)
564
# Not passing a progress bar, because it creates a new
565
# progress, which overwrites the current progress,
566
# and doesn't look nice
568
checker.checked_weaves[self.file_id] = True
570
w = tree.get_weave(self.file_id)
572
mutter('check version {%s} of {%s}', tree_revision_id, self.file_id)
573
checker.checked_text_cnt += 1
574
# We can't check the length, because Weave doesn't store that
575
# information, and the whole point of looking at the weave's
576
# sha1sum is that we don't have to extract the text.
577
if self.text_sha1 != w.get_sha1(self.revision):
578
raise BzrCheckError('text {%s} version {%s} wrong sha1'
579
% (self.file_id, self.revision))
580
checker.checked_texts[t] = self.text_sha1
583
other = InventoryFile(self.file_id, self.name, self.parent_id)
584
other.executable = self.executable
585
other.text_id = self.text_id
586
other.text_sha1 = self.text_sha1
587
other.text_size = self.text_size
588
other.revision = self.revision
591
def detect_changes(self, old_entry):
592
"""See InventoryEntry.detect_changes."""
593
assert self.text_sha1 != None
594
assert old_entry.text_sha1 != None
595
text_modified = (self.text_sha1 != old_entry.text_sha1)
596
meta_modified = (self.executable != old_entry.executable)
597
return text_modified, meta_modified
599
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
600
output_to, reverse=False):
601
"""See InventoryEntry._diff."""
603
from_text = tree.get_file(self.file_id).readlines()
605
to_text = to_tree.get_file(to_entry.file_id).readlines()
609
text_diff(from_label, from_text,
610
to_label, to_text, output_to)
612
text_diff(to_label, to_text,
613
from_label, from_text, output_to)
616
label_pair = (to_label, from_label)
618
label_pair = (from_label, to_label)
619
print >> output_to, "Binary files %s and %s differ" % label_pair
622
"""See InventoryEntry.has_text."""
625
def __init__(self, file_id, name, parent_id):
626
super(InventoryFile, self).__init__(file_id, name, parent_id)
629
def kind_character(self):
630
"""See InventoryEntry.kind_character."""
633
def _put_in_tar(self, item, tree):
634
"""See InventoryEntry._put_in_tar."""
635
item.type = tarfile.REGTYPE
636
fileobj = tree.get_file(self.file_id)
637
item.size = self.text_size
638
if tree.is_executable(self.file_id):
644
def _put_on_disk(self, fullpath, tree):
645
"""See InventoryEntry._put_on_disk."""
646
pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
647
if tree.is_executable(self.file_id):
648
os.chmod(fullpath, 0755)
650
def _read_tree_state(self, path, work_tree):
651
"""See InventoryEntry._read_tree_state."""
652
self.text_sha1 = work_tree.get_file_sha1(self.file_id, path=path)
653
self.executable = work_tree.is_executable(self.file_id, path=path)
655
def _forget_tree_state(self):
656
self.text_sha1 = None
657
self.executable = None
659
def _snapshot_text(self, file_parents, work_tree, versionedfile_store, transaction):
660
"""See InventoryEntry._snapshot_text."""
661
mutter('storing text of file {%s} in revision {%s} into %r',
662
self.file_id, self.revision, versionedfile_store)
663
# special case to avoid diffing on renames or
665
if (len(file_parents) == 1
666
and self.text_sha1 == file_parents.values()[0].text_sha1
667
and self.text_size == file_parents.values()[0].text_size):
668
previous_ie = file_parents.values()[0]
669
versionedfile = versionedfile_store.get_weave(self.file_id, transaction)
670
versionedfile.clone_text(self.revision, previous_ie.revision, file_parents.keys())
672
new_lines = work_tree.get_file(self.file_id).readlines()
673
self._add_text_to_weave(new_lines, file_parents.keys(), versionedfile_store,
675
self.text_sha1 = sha_strings(new_lines)
676
self.text_size = sum(map(len, new_lines))
679
def _unchanged(self, previous_ie):
680
"""See InventoryEntry._unchanged."""
681
compatible = super(InventoryFile, self)._unchanged(previous_ie)
682
if self.text_sha1 != previous_ie.text_sha1:
685
# FIXME: 20050930 probe for the text size when getting sha1
686
# in _read_tree_state
687
self.text_size = previous_ie.text_size
688
if self.executable != previous_ie.executable:
693
class InventoryLink(InventoryEntry):
694
"""A file in an inventory."""
696
__slots__ = ['symlink_target']
698
def _check(self, checker, rev_id, tree):
699
"""See InventoryEntry._check"""
700
if self.text_sha1 != None or self.text_size != None or self.text_id != None:
701
raise BzrCheckError('symlink {%s} has text in revision {%s}'
702
% (self.file_id, rev_id))
703
if self.symlink_target == None:
704
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
705
% (self.file_id, rev_id))
708
other = InventoryLink(self.file_id, self.name, self.parent_id)
709
other.symlink_target = self.symlink_target
710
other.revision = self.revision
713
def detect_changes(self, old_entry):
714
"""See InventoryEntry.detect_changes."""
715
# FIXME: which _modified field should we use ? RBC 20051003
716
text_modified = (self.symlink_target != old_entry.symlink_target)
718
mutter(" symlink target changed")
719
meta_modified = False
720
return text_modified, meta_modified
722
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
723
output_to, reverse=False):
724
"""See InventoryEntry._diff."""
725
from_text = self.symlink_target
726
if to_entry is not None:
727
to_text = to_entry.symlink_target
732
print >>output_to, '=== target changed %r => %r' % (from_text, to_text)
735
print >>output_to, '=== target was %r' % self.symlink_target
737
print >>output_to, '=== target is %r' % self.symlink_target
739
def __init__(self, file_id, name, parent_id):
740
super(InventoryLink, self).__init__(file_id, name, parent_id)
741
self.kind = 'symlink'
743
def kind_character(self):
744
"""See InventoryEntry.kind_character."""
747
def _put_in_tar(self, item, tree):
748
"""See InventoryEntry._put_in_tar."""
749
item.type = tarfile.SYMTYPE
753
item.linkname = self.symlink_target
756
def _put_on_disk(self, fullpath, tree):
757
"""See InventoryEntry._put_on_disk."""
759
os.symlink(self.symlink_target, fullpath)
761
raise BzrError("Failed to create symlink %r -> %r, error: %s" % (fullpath, self.symlink_target, e))
763
def _read_tree_state(self, path, work_tree):
764
"""See InventoryEntry._read_tree_state."""
765
self.symlink_target = work_tree.get_symlink_target(self.file_id)
767
def _forget_tree_state(self):
768
self.symlink_target = None
770
def _unchanged(self, previous_ie):
771
"""See InventoryEntry._unchanged."""
772
compatible = super(InventoryLink, self)._unchanged(previous_ie)
773
if self.symlink_target != previous_ie.symlink_target:
778
class Inventory(object):
779
"""Inventory of versioned files in a tree.
781
This describes which file_id is present at each point in the tree,
782
and possibly the SHA-1 or other information about the file.
783
Entries can be looked up either by path or by file_id.
785
The inventory represents a typical unix file tree, with
786
directories containing files and subdirectories. We never store
787
the full path to a file, because renaming a directory implicitly
788
moves all of its contents. This class internally maintains a
789
lookup tree that allows the children under a directory to be
792
InventoryEntry objects must not be modified after they are
793
inserted, other than through the Inventory API.
795
>>> inv = Inventory()
796
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
797
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT')
798
>>> inv['123-123'].name
801
May be treated as an iterator or set to look up file ids:
803
>>> bool(inv.path2id('hello.c'))
808
May also look up by name:
810
>>> [x[0] for x in inv.iter_entries()]
812
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
813
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
814
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT-12345678-12345678')
816
def __init__(self, root_id=ROOT_ID, revision_id=None):
817
"""Create or read an inventory.
819
If a working directory is specified, the inventory is read
820
from there. If the file is specified, read from that. If not,
821
the inventory is created empty.
823
The inventory is created with a default root directory, with
826
if root_id is not None:
827
self.root = InventoryDirectory(root_id, '', None)
828
self._byid = {self.root.file_id: self.root}
832
self.revision_id = revision_id
836
# TODO: jam 20051218 Should copy also copy the revision_id?
837
other = Inventory(self.root.file_id)
838
# copy recursively so we know directories will be added before
839
# their children. There are more efficient ways than this...
840
for path, entry in self.iter_entries():
841
if entry == self.root:
843
other.add(entry.copy())
848
return iter(self._byid)
852
"""Returns number of entries."""
853
return len(self._byid)
856
def iter_entries(self, from_dir=None):
857
"""Return (path, entry) pairs, in order by name."""
859
if self.root is None:
862
elif isinstance(from_dir, basestring):
863
from_dir = self._byid[from_dir]
865
# unrolling the recursive called changed the time from
866
# 440ms/663ms (inline/total) to 116ms/116ms
867
children = from_dir.children.items()
869
children = collections.deque(children)
870
stack = [(u'', children)]
872
from_dir_relpath, children = stack[-1]
875
name, ie = children.popleft()
877
# we know that from_dir_relpath never ends in a slash
878
# and 'f' doesn't begin with one, we can do a string op, rather
879
# than the checks of pathjoin(), though this means that all paths
881
path = from_dir_relpath + '/' + name
885
if ie.kind != 'directory':
888
# But do this child first
889
new_children = ie.children.items()
891
new_children = collections.deque(new_children)
892
stack.append((path, new_children))
893
# Break out of inner loop, so that we start outer loop with child
896
# if we finished all children, pop it off the stack
900
"""Return list of (path, ie) for all entries except the root.
902
This may be faster than iter_entries.
905
def descend(dir_ie, dir_path):
906
kids = dir_ie.children.items()
908
for name, ie in kids:
909
child_path = pathjoin(dir_path, name)
910
accum.append((child_path, ie))
911
if ie.kind == 'directory':
912
descend(ie, child_path)
914
descend(self.root, u'')
918
def directories(self):
919
"""Return (path, entry) pairs for all directories, including the root.
922
def descend(parent_ie, parent_path):
923
accum.append((parent_path, parent_ie))
925
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
928
for name, child_ie in kids:
929
child_path = pathjoin(parent_path, name)
930
descend(child_ie, child_path)
931
descend(self.root, u'')
936
def __contains__(self, file_id):
937
"""True if this entry contains a file with given id.
939
>>> inv = Inventory()
940
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
941
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT')
947
return file_id in self._byid
950
def __getitem__(self, file_id):
951
"""Return the entry for given file_id.
953
>>> inv = Inventory()
954
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
955
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT')
956
>>> inv['123123'].name
960
return self._byid[file_id]
963
raise BzrError("can't look up file_id None")
965
raise BzrError("file_id {%s} not in inventory" % file_id)
968
def get_file_kind(self, file_id):
969
return self._byid[file_id].kind
971
def get_child(self, parent_id, filename):
972
return self[parent_id].children.get(filename)
975
def add(self, entry):
976
"""Add entry to inventory.
978
To add a file to a branch ready to be committed, use Branch.add,
981
Returns the new entry object.
983
if entry.file_id in self._byid:
984
raise BzrError("inventory already contains entry with id {%s}" % entry.file_id)
986
if entry.parent_id == ROOT_ID or entry.parent_id is None:
987
entry.parent_id = self.root.file_id
990
parent = self._byid[entry.parent_id]
992
raise BzrError("parent_id {%s} not in inventory" % entry.parent_id)
994
if parent.children.has_key(entry.name):
995
raise BzrError("%s is already versioned" %
996
pathjoin(self.id2path(parent.file_id), entry.name))
998
self._byid[entry.file_id] = entry
999
parent.children[entry.name] = entry
1003
def add_path(self, relpath, kind, file_id=None, parent_id=None):
1004
"""Add entry from a path.
1006
The immediate parent must already be versioned.
1008
Returns the new entry object."""
1010
parts = bzrlib.osutils.splitpath(relpath)
1014
file_id = bzrlib.workingtree.gen_root_id()
1015
self.root = InventoryDirectory(file_id, '', None)
1016
self._byid = {self.root.file_id: self.root}
1019
parent_path = parts[:-1]
1020
parent_id = self.path2id(parent_path)
1021
if parent_id == None:
1022
raise NotVersionedError(path=parent_path)
1023
ie = make_entry(kind, parts[-1], parent_id, file_id)
1026
def __delitem__(self, file_id):
1027
"""Remove entry by id.
1029
>>> inv = Inventory()
1030
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1031
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT')
1040
assert ie.parent_id is None or \
1041
self[ie.parent_id].children[ie.name] == ie
1043
del self._byid[file_id]
1044
if ie.parent_id is not None:
1045
del self[ie.parent_id].children[ie.name]
1048
def __eq__(self, other):
1049
"""Compare two sets by comparing their contents.
1051
>>> i1 = Inventory()
1052
>>> i2 = Inventory()
1055
>>> i1.add(InventoryFile('123', 'foo', ROOT_ID))
1056
InventoryFile('123', 'foo', parent_id='TREE_ROOT')
1059
>>> i2.add(InventoryFile('123', 'foo', ROOT_ID))
1060
InventoryFile('123', 'foo', parent_id='TREE_ROOT')
1064
if not isinstance(other, Inventory):
1065
return NotImplemented
1067
if len(self._byid) != len(other._byid):
1068
# shortcut: obviously not the same
1071
return self._byid == other._byid
1074
def __ne__(self, other):
1075
return not self.__eq__(other)
1079
raise ValueError('not hashable')
1081
def _iter_file_id_parents(self, file_id):
1082
"""Yield the parents of file_id up to the root."""
1083
while file_id != None:
1085
ie = self._byid[file_id]
1087
raise BzrError("file_id {%s} not found in inventory" % file_id)
1089
file_id = ie.parent_id
1091
def get_idpath(self, file_id):
1092
"""Return a list of file_ids for the path to an entry.
1094
The list contains one element for each directory followed by
1095
the id of the file itself. So the length of the returned list
1096
is equal to the depth of the file in the tree, counting the
1097
root directory as depth 1.
1100
for parent in self._iter_file_id_parents(file_id):
1101
p.insert(0, parent.file_id)
1104
def id2path(self, file_id):
1105
"""Return as a string the path to file_id.
1108
>>> e = i.add(InventoryDirectory('src-id', 'src', ROOT_ID))
1109
>>> e = i.add(InventoryFile('foo-id', 'foo.c', parent_id='src-id'))
1110
>>> print i.id2path('foo-id')
1113
# get all names, skipping root
1114
return '/'.join(reversed(
1115
[parent.name for parent in
1116
self._iter_file_id_parents(file_id)][:-1]))
1118
def path2id(self, name):
1119
"""Walk down through directories to return entry of last component.
1121
names may be either a list of path components, or a single
1122
string, in which case it is automatically split.
1124
This returns the entry of the last component in the path,
1125
which may be either a file or a directory.
1127
Returns None IFF the path is not found.
1129
if isinstance(name, types.StringTypes):
1130
name = splitpath(name)
1132
# mutter("lookup path %r" % name)
1139
cie = parent.children[f]
1140
assert cie.name == f
1141
assert cie.parent_id == parent.file_id
1144
# or raise an error?
1147
return parent.file_id
1150
def has_filename(self, names):
1151
return bool(self.path2id(names))
1154
def has_id(self, file_id):
1155
return self._byid.has_key(file_id)
1158
def rename(self, file_id, new_parent_id, new_name):
1159
"""Move a file within the inventory.
1161
This can change either the name, or the parent, or both.
1163
This does not move the working file."""
1164
if not is_valid_name(new_name):
1165
raise BzrError("not an acceptable filename: %r" % new_name)
1167
new_parent = self._byid[new_parent_id]
1168
if new_name in new_parent.children:
1169
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
1171
new_parent_idpath = self.get_idpath(new_parent_id)
1172
if file_id in new_parent_idpath:
1173
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
1174
% (self.id2path(file_id), self.id2path(new_parent_id)))
1176
file_ie = self._byid[file_id]
1177
old_parent = self._byid[file_ie.parent_id]
1179
# TODO: Don't leave things messed up if this fails
1181
del old_parent.children[file_ie.name]
1182
new_parent.children[new_name] = file_ie
1184
file_ie.name = new_name
1185
file_ie.parent_id = new_parent_id
1187
def is_root(self, file_id):
1188
return self.root is not None and file_id == self.root.file_id
1191
def make_entry(kind, name, parent_id, file_id=None):
1192
"""Create an inventory entry.
1194
:param kind: the type of inventory entry to create.
1195
:param name: the basename of the entry.
1196
:param parent_id: the parent_id of the entry.
1197
:param file_id: the file_id to use. if None, one will be created.
1200
file_id = bzrlib.workingtree.gen_file_id(name)
1201
if kind == 'directory':
1202
return InventoryDirectory(file_id, name, parent_id)
1203
elif kind == 'file':
1204
return InventoryFile(file_id, name, parent_id)
1205
elif kind == 'symlink':
1206
return InventoryLink(file_id, name, parent_id)
1208
raise BzrError("unknown kind %r" % kind)
1214
def is_valid_name(name):
1216
if _NAME_RE == None:
1217
_NAME_RE = re.compile(r'^[^/\\]+$')
1219
return bool(_NAME_RE.match(name))
added
removed
bzrlib/inventory.py
