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.
34
from bzrlib.lazy_import import lazy_import
35
lazy_import(globals(), """
49
from bzrlib.errors import (
53
from bzrlib.trace import mutter
56
class InventoryEntry(object):
57
"""Description of a versioned file.
59
An InventoryEntry has the following fields, which are also
60
present in the XML inventory-entry element:
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.)
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])
97
(True, InventoryDirectory('TREE_ROOT', '', 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('2323', 'bye.c', '123'))
101
Traceback (most recent call last):
103
BzrError: inventory already contains entry with id {2323}
104
>>> i.add(InventoryFile('2324', 'bye.c', '123'))
105
InventoryFile('2324', 'bye.c', parent_id='123', sha1=None, len=None)
106
>>> i.add(InventoryDirectory('2325', 'wibble', '123'))
107
InventoryDirectory('2325', 'wibble', parent_id='123', revision=None)
108
>>> i.path2id('src/wibble')
112
>>> i.add(InventoryFile('2326', 'wibble.c', '2325'))
113
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
115
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
116
>>> for path, entry in i.iter_entries():
118
... assert i.path2id(path)
126
>>> i.id2path('2326')
127
'src/wibble/wibble.c'
130
# Constants returned by describe_change()
132
# TODO: These should probably move to some kind of FileChangeDescription
133
# class; that's like what's inside a TreeDelta but we want to be able to
134
# generate them just for one file at a time.
136
MODIFIED_AND_RENAMED = 'modified and renamed'
140
def detect_changes(self, old_entry):
141
"""Return a (text_modified, meta_modified) from this to old_entry.
143
_read_tree_state must have been called on self and old_entry prior to
144
calling detect_changes.
148
def diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
149
output_to, reverse=False):
150
"""Perform a diff from this to to_entry.
152
text_diff will be used for textual difference calculation.
153
This is a template method, override _diff in child classes.
155
self._read_tree_state(tree.id2path(self.file_id), tree)
157
# cannot diff from one kind to another - you must do a removal
158
# and an addif they do not match.
159
assert self.kind == to_entry.kind
160
to_entry._read_tree_state(to_tree.id2path(to_entry.file_id),
162
self._diff(text_diff, from_label, tree, to_label, to_entry, to_tree,
165
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
166
output_to, reverse=False):
167
"""Perform a diff between two entries of the same kind."""
169
def find_previous_heads(self, previous_inventories,
170
versioned_file_store,
173
"""Return the revisions and entries that directly precede this.
175
Returned as a map from revision to inventory entry.
177
This is a map containing the file revisions in all parents
178
for which the file exists, and its revision is not a parent of
179
any other. If the file is new, the set will be empty.
181
:param versioned_file_store: A store where ancestry data on this
182
file id can be queried.
183
:param transaction: The transaction that queries to the versioned
184
file store should be completed under.
185
:param entry_vf: The entry versioned file, if its already available.
187
def get_ancestors(weave, entry):
188
return set(weave.get_ancestry(entry.revision))
189
# revision:ie mapping for each ie found in previous_inventories.
191
# revision:ie mapping with one revision for each head.
193
# revision: ancestor list for each head
195
# identify candidate head revision ids.
196
for inv in previous_inventories:
197
if self.file_id in inv:
198
ie = inv[self.file_id]
199
assert ie.file_id == self.file_id
200
if ie.revision in candidates:
201
# same revision value in two different inventories:
202
# correct possible inconsistencies:
203
# * there was a bug in revision updates with 'x' bit
206
if candidates[ie.revision].executable != ie.executable:
207
candidates[ie.revision].executable = False
208
ie.executable = False
209
except AttributeError:
211
# must now be the same.
212
assert candidates[ie.revision] == ie
214
# add this revision as a candidate.
215
candidates[ie.revision] = ie
217
# common case optimisation
218
if len(candidates) == 1:
219
# if there is only one candidate revision found
220
# then we can opening the versioned file to access ancestry:
221
# there cannot be any ancestors to eliminate when there is
222
# only one revision available.
223
heads[ie.revision] = ie
226
# eliminate ancestors amongst the available candidates:
227
# heads are those that are not an ancestor of any other candidate
228
# - this provides convergence at a per-file level.
229
for ie in candidates.values():
230
# may be an ancestor of a known head:
231
already_present = 0 != len(
232
[head for head in heads
233
if ie.revision in head_ancestors[head]])
235
# an ancestor of an analyzed candidate.
237
# not an ancestor of a known head:
238
# load the versioned file for this file id if needed
240
entry_vf = versioned_file_store.get_weave_or_empty(
241
self.file_id, transaction)
242
ancestors = get_ancestors(entry_vf, ie)
243
# may knock something else out:
244
check_heads = list(heads.keys())
245
for head in check_heads:
246
if head in ancestors:
247
# this previously discovered 'head' is not
248
# really a head - its an ancestor of the newly
251
head_ancestors[ie.revision] = ancestors
252
heads[ie.revision] = ie
255
def get_tar_item(self, root, dp, now, tree):
256
"""Get a tarfile item and a file stream for its content."""
257
item = tarfile.TarInfo(osutils.pathjoin(root, dp).encode('utf8'))
258
# TODO: would be cool to actually set it to the timestamp of the
259
# revision it was last changed
261
fileobj = self._put_in_tar(item, tree)
265
"""Return true if the object this entry represents has textual data.
267
Note that textual data includes binary content.
269
Also note that all entries get weave files created for them.
270
This attribute is primarily used when upgrading from old trees that
271
did not have the weave index for all inventory entries.
275
def __init__(self, file_id, name, parent_id, text_id=None):
276
"""Create an InventoryEntry
278
The filename must be a single component, relative to the
279
parent directory; it cannot be a whole path or relative name.
281
>>> e = InventoryFile('123', 'hello.c', ROOT_ID)
286
>>> e = InventoryFile('123', 'src/hello.c', ROOT_ID)
287
Traceback (most recent call last):
288
InvalidEntryName: Invalid entry name: src/hello.c
290
assert isinstance(name, basestring), name
291
if '/' in name or '\\' in name:
292
raise errors.InvalidEntryName(name=name)
293
self.executable = False
295
self.text_sha1 = None
296
self.text_size = None
297
self.file_id = file_id
299
self.text_id = text_id
300
self.parent_id = parent_id
301
self.symlink_target = None
303
def kind_character(self):
304
"""Return a short kind indicator useful for appending to names."""
305
raise BzrError('unknown kind %r' % self.kind)
307
known_kinds = ('file', 'directory', 'symlink')
309
def _put_in_tar(self, item, tree):
310
"""populate item for stashing in a tar, and return the content stream.
312
If no content is available, return None.
314
raise BzrError("don't know how to export {%s} of kind %r" %
315
(self.file_id, self.kind))
317
def put_on_disk(self, dest, dp, tree):
318
"""Create a representation of self on disk in the prefix dest.
320
This is a template method - implement _put_on_disk in subclasses.
322
fullpath = osutils.pathjoin(dest, dp)
323
self._put_on_disk(fullpath, tree)
324
# mutter(" export {%s} kind %s to %s", self.file_id,
325
# self.kind, fullpath)
327
def _put_on_disk(self, fullpath, tree):
328
"""Put this entry onto disk at fullpath, from tree tree."""
329
raise BzrError("don't know how to export {%s} of kind %r" % (self.file_id, self.kind))
331
def sorted_children(self):
332
return sorted(self.children.items())
335
def versionable_kind(kind):
336
return (kind in ('file', 'directory', 'symlink'))
338
def check(self, checker, rev_id, inv, tree):
339
"""Check this inventory entry is intact.
341
This is a template method, override _check for kind specific
344
:param checker: Check object providing context for the checks;
345
can be used to find out what parts of the repository have already
347
:param rev_id: Revision id from which this InventoryEntry was loaded.
348
Not necessarily the last-changed revision for this file.
349
:param inv: Inventory from which the entry was loaded.
350
:param tree: RevisionTree for this entry.
352
if self.parent_id is not None:
353
if not inv.has_id(self.parent_id):
354
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
355
% (self.parent_id, rev_id))
356
self._check(checker, rev_id, tree)
358
def _check(self, checker, rev_id, tree):
359
"""Check this inventory entry for kind specific errors."""
360
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
364
"""Clone this inventory entry."""
365
raise NotImplementedError
368
def describe_change(old_entry, new_entry):
369
"""Describe the change between old_entry and this.
371
This smells of being an InterInventoryEntry situation, but as its
372
the first one, we're making it a static method for now.
374
An entry with a different parent, or different name is considered
375
to be renamed. Reparenting is an internal detail.
376
Note that renaming the parent does not trigger a rename for the
379
# TODO: Perhaps return an object rather than just a string
380
if old_entry is new_entry:
381
# also the case of both being None
383
elif old_entry is None:
385
elif new_entry is None:
387
if old_entry.kind != new_entry.kind:
389
text_modified, meta_modified = new_entry.detect_changes(old_entry)
390
if text_modified or meta_modified:
394
# TODO 20060511 (mbp, rbc) factor out 'detect_rename' here.
395
if old_entry.parent_id != new_entry.parent_id:
397
elif old_entry.name != new_entry.name:
401
if renamed and not modified:
402
return InventoryEntry.RENAMED
403
if modified and not renamed:
405
if modified and renamed:
406
return InventoryEntry.MODIFIED_AND_RENAMED
410
return ("%s(%r, %r, parent_id=%r, revision=%r)"
411
% (self.__class__.__name__,
417
def snapshot(self, revision, path, previous_entries,
418
work_tree, commit_builder):
419
"""Make a snapshot of this entry which may or may not have changed.
421
This means that all its fields are populated, that it has its
422
text stored in the text store or weave.
424
# mutter('new parents of %s are %r', path, previous_entries)
425
self._read_tree_state(path, work_tree)
426
# TODO: Where should we determine whether to reuse a
427
# previous revision id or create a new revision? 20060606
428
if len(previous_entries) == 1:
429
# cannot be unchanged unless there is only one parent file rev.
430
parent_ie = previous_entries.values()[0]
431
if self._unchanged(parent_ie):
432
# mutter("found unchanged entry")
433
self.revision = parent_ie.revision
435
return self._snapshot_into_revision(revision, previous_entries,
436
work_tree, commit_builder)
438
def _snapshot_into_revision(self, revision, previous_entries, work_tree,
440
"""Record this revision unconditionally into a store.
442
The entry's last-changed revision property (`revision`) is updated to
443
that of the new revision.
445
:param revision: id of the new revision that is being recorded.
447
:returns: String description of the commit (e.g. "merged", "modified"), etc.
449
# mutter('new revision {%s} for {%s}', revision, self.file_id)
450
self.revision = revision
451
self._snapshot_text(previous_entries, work_tree, commit_builder)
453
def _snapshot_text(self, file_parents, work_tree, commit_builder):
454
"""Record the 'text' of this entry, whatever form that takes.
456
This default implementation simply adds an empty text.
458
raise NotImplementedError(self._snapshot_text)
460
def __eq__(self, other):
461
if not isinstance(other, InventoryEntry):
462
return NotImplemented
464
return ((self.file_id == other.file_id)
465
and (self.name == other.name)
466
and (other.symlink_target == self.symlink_target)
467
and (self.text_sha1 == other.text_sha1)
468
and (self.text_size == other.text_size)
469
and (self.text_id == other.text_id)
470
and (self.parent_id == other.parent_id)
471
and (self.kind == other.kind)
472
and (self.revision == other.revision)
473
and (self.executable == other.executable)
476
def __ne__(self, other):
477
return not (self == other)
480
raise ValueError('not hashable')
482
def _unchanged(self, previous_ie):
483
"""Has this entry changed relative to previous_ie.
485
This method should be overridden in child classes.
488
# different inv parent
489
if previous_ie.parent_id != self.parent_id:
492
elif previous_ie.name != self.name:
496
def _read_tree_state(self, path, work_tree):
497
"""Populate fields in the inventory entry from the given tree.
499
Note that this should be modified to be a noop on virtual trees
500
as all entries created there are prepopulated.
502
# TODO: Rather than running this manually, we should check the
503
# working sha1 and other expensive properties when they're
504
# first requested, or preload them if they're already known
505
pass # nothing to do by default
507
def _forget_tree_state(self):
511
class RootEntry(InventoryEntry):
513
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
514
'text_id', 'parent_id', 'children', 'executable',
515
'revision', 'symlink_target']
517
def _check(self, checker, rev_id, tree):
518
"""See InventoryEntry._check"""
520
def __init__(self, file_id):
521
self.file_id = file_id
523
self.kind = 'directory'
524
self.parent_id = None
527
symbol_versioning.warn('RootEntry is deprecated as of bzr 0.10.'
528
' Please use InventoryDirectory instead.',
529
DeprecationWarning, stacklevel=2)
531
def __eq__(self, other):
532
if not isinstance(other, RootEntry):
533
return NotImplemented
535
return (self.file_id == other.file_id) \
536
and (self.children == other.children)
539
class InventoryDirectory(InventoryEntry):
540
"""A directory in an inventory."""
542
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
543
'text_id', 'parent_id', 'children', 'executable',
544
'revision', 'symlink_target']
546
def _check(self, checker, rev_id, tree):
547
"""See InventoryEntry._check"""
548
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
549
raise BzrCheckError('directory {%s} has text in revision {%s}'
550
% (self.file_id, rev_id))
553
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
554
other.revision = self.revision
555
# note that children are *not* copied; they're pulled across when
559
def __init__(self, file_id, name, parent_id):
560
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
562
self.kind = 'directory'
564
def kind_character(self):
565
"""See InventoryEntry.kind_character."""
568
def _put_in_tar(self, item, tree):
569
"""See InventoryEntry._put_in_tar."""
570
item.type = tarfile.DIRTYPE
577
def _put_on_disk(self, fullpath, tree):
578
"""See InventoryEntry._put_on_disk."""
581
def _snapshot_text(self, file_parents, work_tree, commit_builder):
582
"""See InventoryEntry._snapshot_text."""
583
commit_builder.modified_directory(self.file_id, file_parents)
586
class InventoryFile(InventoryEntry):
587
"""A file in an inventory."""
589
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
590
'text_id', 'parent_id', 'children', 'executable',
591
'revision', 'symlink_target']
593
def _check(self, checker, tree_revision_id, tree):
594
"""See InventoryEntry._check"""
595
t = (self.file_id, self.revision)
596
if t in checker.checked_texts:
597
prev_sha = checker.checked_texts[t]
598
if prev_sha != self.text_sha1:
599
raise BzrCheckError('mismatched sha1 on {%s} in {%s}' %
600
(self.file_id, tree_revision_id))
602
checker.repeated_text_cnt += 1
605
if self.file_id not in checker.checked_weaves:
606
mutter('check weave {%s}', self.file_id)
607
w = tree.get_weave(self.file_id)
608
# Not passing a progress bar, because it creates a new
609
# progress, which overwrites the current progress,
610
# and doesn't look nice
612
checker.checked_weaves[self.file_id] = True
614
w = tree.get_weave(self.file_id)
616
mutter('check version {%s} of {%s}', tree_revision_id, self.file_id)
617
checker.checked_text_cnt += 1
618
# We can't check the length, because Weave doesn't store that
619
# information, and the whole point of looking at the weave's
620
# sha1sum is that we don't have to extract the text.
621
if self.text_sha1 != w.get_sha1(self.revision):
622
raise BzrCheckError('text {%s} version {%s} wrong sha1'
623
% (self.file_id, self.revision))
624
checker.checked_texts[t] = self.text_sha1
627
other = InventoryFile(self.file_id, self.name, self.parent_id)
628
other.executable = self.executable
629
other.text_id = self.text_id
630
other.text_sha1 = self.text_sha1
631
other.text_size = self.text_size
632
other.revision = self.revision
635
def detect_changes(self, old_entry):
636
"""See InventoryEntry.detect_changes."""
637
assert self.text_sha1 is not None
638
assert old_entry.text_sha1 is not None
639
text_modified = (self.text_sha1 != old_entry.text_sha1)
640
meta_modified = (self.executable != old_entry.executable)
641
return text_modified, meta_modified
643
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
644
output_to, reverse=False):
645
"""See InventoryEntry._diff."""
647
from_text = tree.get_file(self.file_id).readlines()
649
to_text = to_tree.get_file(to_entry.file_id).readlines()
653
text_diff(from_label, from_text,
654
to_label, to_text, output_to)
656
text_diff(to_label, to_text,
657
from_label, from_text, output_to)
658
except errors.BinaryFile:
660
label_pair = (to_label, from_label)
662
label_pair = (from_label, to_label)
663
print >> output_to, "Binary files %s and %s differ" % label_pair
666
"""See InventoryEntry.has_text."""
669
def __init__(self, file_id, name, parent_id):
670
super(InventoryFile, self).__init__(file_id, name, parent_id)
673
def kind_character(self):
674
"""See InventoryEntry.kind_character."""
677
def _put_in_tar(self, item, tree):
678
"""See InventoryEntry._put_in_tar."""
679
item.type = tarfile.REGTYPE
680
fileobj = tree.get_file(self.file_id)
681
item.size = self.text_size
682
if tree.is_executable(self.file_id):
688
def _put_on_disk(self, fullpath, tree):
689
"""See InventoryEntry._put_on_disk."""
690
osutils.pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
691
if tree.is_executable(self.file_id):
692
os.chmod(fullpath, 0755)
694
def _read_tree_state(self, path, work_tree):
695
"""See InventoryEntry._read_tree_state."""
696
self.text_sha1 = work_tree.get_file_sha1(self.file_id, path=path)
697
# FIXME: 20050930 probe for the text size when getting sha1
698
# in _read_tree_state
699
self.executable = work_tree.is_executable(self.file_id, path=path)
702
return ("%s(%r, %r, parent_id=%r, sha1=%r, len=%s)"
703
% (self.__class__.__name__,
710
def _forget_tree_state(self):
711
self.text_sha1 = None
713
def _snapshot_text(self, file_parents, work_tree, commit_builder):
714
"""See InventoryEntry._snapshot_text."""
715
def get_content_byte_lines():
716
return work_tree.get_file(self.file_id).readlines()
717
self.text_sha1, self.text_size = commit_builder.modified_file_text(
718
self.file_id, file_parents, get_content_byte_lines, self.text_sha1, self.text_size)
720
def _unchanged(self, previous_ie):
721
"""See InventoryEntry._unchanged."""
722
compatible = super(InventoryFile, self)._unchanged(previous_ie)
723
if self.text_sha1 != previous_ie.text_sha1:
726
# FIXME: 20050930 probe for the text size when getting sha1
727
# in _read_tree_state
728
self.text_size = previous_ie.text_size
729
if self.executable != previous_ie.executable:
734
class InventoryLink(InventoryEntry):
735
"""A file in an inventory."""
737
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
738
'text_id', 'parent_id', 'children', 'executable',
739
'revision', 'symlink_target']
741
def _check(self, checker, rev_id, tree):
742
"""See InventoryEntry._check"""
743
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
744
raise BzrCheckError('symlink {%s} has text in revision {%s}'
745
% (self.file_id, rev_id))
746
if self.symlink_target is None:
747
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
748
% (self.file_id, rev_id))
751
other = InventoryLink(self.file_id, self.name, self.parent_id)
752
other.symlink_target = self.symlink_target
753
other.revision = self.revision
756
def detect_changes(self, old_entry):
757
"""See InventoryEntry.detect_changes."""
758
# FIXME: which _modified field should we use ? RBC 20051003
759
text_modified = (self.symlink_target != old_entry.symlink_target)
761
mutter(" symlink target changed")
762
meta_modified = False
763
return text_modified, meta_modified
765
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
766
output_to, reverse=False):
767
"""See InventoryEntry._diff."""
768
from_text = self.symlink_target
769
if to_entry is not None:
770
to_text = to_entry.symlink_target
775
print >>output_to, '=== target changed %r => %r' % (from_text, to_text)
778
print >>output_to, '=== target was %r' % self.symlink_target
780
print >>output_to, '=== target is %r' % self.symlink_target
782
def __init__(self, file_id, name, parent_id):
783
super(InventoryLink, self).__init__(file_id, name, parent_id)
784
self.kind = 'symlink'
786
def kind_character(self):
787
"""See InventoryEntry.kind_character."""
790
def _put_in_tar(self, item, tree):
791
"""See InventoryEntry._put_in_tar."""
792
item.type = tarfile.SYMTYPE
796
item.linkname = self.symlink_target
799
def _put_on_disk(self, fullpath, tree):
800
"""See InventoryEntry._put_on_disk."""
802
os.symlink(self.symlink_target, fullpath)
804
raise BzrError("Failed to create symlink %r -> %r, error: %s" % (fullpath, self.symlink_target, e))
806
def _read_tree_state(self, path, work_tree):
807
"""See InventoryEntry._read_tree_state."""
808
self.symlink_target = work_tree.get_symlink_target(self.file_id)
810
def _forget_tree_state(self):
811
self.symlink_target = None
813
def _unchanged(self, previous_ie):
814
"""See InventoryEntry._unchanged."""
815
compatible = super(InventoryLink, self)._unchanged(previous_ie)
816
if self.symlink_target != previous_ie.symlink_target:
820
def _snapshot_text(self, file_parents, work_tree, commit_builder):
821
"""See InventoryEntry._snapshot_text."""
822
commit_builder.modified_link(
823
self.file_id, file_parents, self.symlink_target)
826
class Inventory(object):
827
"""Inventory of versioned files in a tree.
829
This describes which file_id is present at each point in the tree,
830
and possibly the SHA-1 or other information about the file.
831
Entries can be looked up either by path or by file_id.
833
The inventory represents a typical unix file tree, with
834
directories containing files and subdirectories. We never store
835
the full path to a file, because renaming a directory implicitly
836
moves all of its contents. This class internally maintains a
837
lookup tree that allows the children under a directory to be
840
InventoryEntry objects must not be modified after they are
841
inserted, other than through the Inventory API.
843
>>> inv = Inventory()
844
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
845
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
846
>>> inv['123-123'].name
849
May be treated as an iterator or set to look up file ids:
851
>>> bool(inv.path2id('hello.c'))
856
May also look up by name:
858
>>> [x[0] for x in inv.iter_entries()]
860
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
861
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
862
Traceback (most recent call last):
863
BzrError: parent_id {TREE_ROOT} not in inventory
864
>>> inv.add(InventoryFile('123-123', 'hello.c', 'TREE_ROOT-12345678-12345678'))
865
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT-12345678-12345678', sha1=None, len=None)
867
def __init__(self, root_id=ROOT_ID, revision_id=None):
868
"""Create or read an inventory.
870
If a working directory is specified, the inventory is read
871
from there. If the file is specified, read from that. If not,
872
the inventory is created empty.
874
The inventory is created with a default root directory, with
877
if root_id is not None:
878
self._set_root(InventoryDirectory(root_id, '', None))
882
self.revision_id = revision_id
884
def _set_root(self, ie):
886
self._byid = {self.root.file_id: self.root}
889
# TODO: jam 20051218 Should copy also copy the revision_id?
890
entries = self.iter_entries()
891
other = Inventory(entries.next()[1].file_id)
892
# copy recursively so we know directories will be added before
893
# their children. There are more efficient ways than this...
894
for path, entry in entries():
895
other.add(entry.copy())
899
return iter(self._byid)
902
"""Returns number of entries."""
903
return len(self._byid)
905
def iter_entries(self, from_dir=None):
906
"""Return (path, entry) pairs, in order by name."""
908
if self.root is None:
912
elif isinstance(from_dir, basestring):
913
from_dir = self._byid[from_dir]
915
# unrolling the recursive called changed the time from
916
# 440ms/663ms (inline/total) to 116ms/116ms
917
children = from_dir.children.items()
919
children = collections.deque(children)
920
stack = [(u'', children)]
922
from_dir_relpath, children = stack[-1]
925
name, ie = children.popleft()
927
# we know that from_dir_relpath never ends in a slash
928
# and 'f' doesn't begin with one, we can do a string op, rather
929
# than the checks of pathjoin(), though this means that all paths
931
path = from_dir_relpath + '/' + name
935
if ie.kind != 'directory':
938
# But do this child first
939
new_children = ie.children.items()
941
new_children = collections.deque(new_children)
942
stack.append((path, new_children))
943
# Break out of inner loop, so that we start outer loop with child
946
# if we finished all children, pop it off the stack
949
def iter_entries_by_dir(self, from_dir=None, specific_file_ids=None):
950
"""Iterate over the entries in a directory first order.
952
This returns all entries for a directory before returning
953
the entries for children of a directory. This is not
954
lexicographically sorted order, and is a hybrid between
955
depth-first and breadth-first.
957
:return: This yields (path, entry) pairs
959
# TODO? Perhaps this should return the from_dir so that the root is
960
# yielded? or maybe an option?
962
if self.root is None:
964
# Optimize a common case
965
if specific_file_ids is not None and len(specific_file_ids) == 1:
966
file_id = list(specific_file_ids)[0]
968
yield self.id2path(file_id), self[file_id]
971
if (specific_file_ids is None or
972
self.root.file_id in specific_file_ids):
974
elif isinstance(from_dir, basestring):
975
from_dir = self._byid[from_dir]
977
if specific_file_ids is not None:
979
def add_ancestors(file_id):
980
if file_id not in self:
982
parent_id = self[file_id].parent_id
983
if parent_id is None:
985
if parent_id not in parents:
986
parents.add(parent_id)
987
add_ancestors(parent_id)
988
for file_id in specific_file_ids:
989
add_ancestors(file_id)
993
stack = [(u'', from_dir)]
995
cur_relpath, cur_dir = stack.pop()
998
for child_name, child_ie in sorted(cur_dir.children.iteritems()):
1000
child_relpath = cur_relpath + child_name
1002
if (specific_file_ids is None or
1003
child_ie.file_id in specific_file_ids):
1004
yield child_relpath, child_ie
1006
if child_ie.kind == 'directory':
1007
if parents is None or child_ie.file_id in parents:
1008
child_dirs.append((child_relpath+'/', child_ie))
1009
stack.extend(reversed(child_dirs))
1012
"""Return list of (path, ie) for all entries except the root.
1014
This may be faster than iter_entries.
1017
def descend(dir_ie, dir_path):
1018
kids = dir_ie.children.items()
1020
for name, ie in kids:
1021
child_path = osutils.pathjoin(dir_path, name)
1022
accum.append((child_path, ie))
1023
if ie.kind == 'directory':
1024
descend(ie, child_path)
1026
descend(self.root, u'')
1029
def directories(self):
1030
"""Return (path, entry) pairs for all directories, including the root.
1033
def descend(parent_ie, parent_path):
1034
accum.append((parent_path, parent_ie))
1036
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
1039
for name, child_ie in kids:
1040
child_path = osutils.pathjoin(parent_path, name)
1041
descend(child_ie, child_path)
1042
descend(self.root, u'')
1045
def __contains__(self, file_id):
1046
"""True if this entry contains a file with given id.
1048
>>> inv = Inventory()
1049
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1050
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
1056
return (file_id in self._byid)
1058
def __getitem__(self, file_id):
1059
"""Return the entry for given file_id.
1061
>>> inv = Inventory()
1062
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
1063
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
1064
>>> inv['123123'].name
1068
return self._byid[file_id]
1070
# really we're passing an inventory, not a tree...
1071
raise errors.NoSuchId(self, file_id)
1073
def get_file_kind(self, file_id):
1074
return self._byid[file_id].kind
1076
def get_child(self, parent_id, filename):
1077
return self[parent_id].children.get(filename)
1079
def add(self, entry):
1080
"""Add entry to inventory.
1082
To add a file to a branch ready to be committed, use Branch.add,
1085
Returns the new entry object.
1087
if entry.file_id in self._byid:
1088
raise BzrError("inventory already contains entry with id {%s}" % entry.file_id)
1090
if entry.parent_id is None:
1091
assert self.root is None and len(self._byid) == 0
1092
self._set_root(entry)
1095
parent = self._byid[entry.parent_id]
1097
raise BzrError("parent_id {%s} not in inventory" % entry.parent_id)
1099
if entry.name in parent.children:
1100
raise BzrError("%s is already versioned" %
1101
osutils.pathjoin(self.id2path(parent.file_id), entry.name))
1103
self._byid[entry.file_id] = entry
1104
parent.children[entry.name] = entry
1107
def add_path(self, relpath, kind, file_id=None, parent_id=None):
1108
"""Add entry from a path.
1110
The immediate parent must already be versioned.
1112
Returns the new entry object."""
1114
parts = osutils.splitpath(relpath)
1118
file_id = generate_ids.gen_root_id()
1119
self.root = InventoryDirectory(file_id, '', None)
1120
self._byid = {self.root.file_id: self.root}
1123
parent_path = parts[:-1]
1124
parent_id = self.path2id(parent_path)
1125
if parent_id is None:
1126
raise errors.NotVersionedError(path=parent_path)
1127
ie = make_entry(kind, parts[-1], parent_id, file_id)
1130
def __delitem__(self, file_id):
1131
"""Remove entry by id.
1133
>>> inv = Inventory()
1134
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1135
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
1144
assert ie.parent_id is None or \
1145
self[ie.parent_id].children[ie.name] == ie
1147
del self._byid[file_id]
1148
if ie.parent_id is not None:
1149
del self[ie.parent_id].children[ie.name]
1151
def __eq__(self, other):
1152
"""Compare two sets by comparing their contents.
1154
>>> i1 = Inventory()
1155
>>> i2 = Inventory()
1158
>>> i1.add(InventoryFile('123', 'foo', ROOT_ID))
1159
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
1162
>>> i2.add(InventoryFile('123', 'foo', ROOT_ID))
1163
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
1167
if not isinstance(other, Inventory):
1168
return NotImplemented
1170
return self._byid == other._byid
1172
def __ne__(self, other):
1173
return not self.__eq__(other)
1176
raise ValueError('not hashable')
1178
def _iter_file_id_parents(self, file_id):
1179
"""Yield the parents of file_id up to the root."""
1180
while file_id is not None:
1182
ie = self._byid[file_id]
1184
raise BzrError("file_id {%s} not found in inventory" % file_id)
1186
file_id = ie.parent_id
1188
def get_idpath(self, file_id):
1189
"""Return a list of file_ids for the path to an entry.
1191
The list contains one element for each directory followed by
1192
the id of the file itself. So the length of the returned list
1193
is equal to the depth of the file in the tree, counting the
1194
root directory as depth 1.
1197
for parent in self._iter_file_id_parents(file_id):
1198
p.insert(0, parent.file_id)
1201
def id2path(self, file_id):
1202
"""Return as a string the path to file_id.
1205
>>> e = i.add(InventoryDirectory('src-id', 'src', ROOT_ID))
1206
>>> e = i.add(InventoryFile('foo-id', 'foo.c', parent_id='src-id'))
1207
>>> print i.id2path('foo-id')
1210
# get all names, skipping root
1211
return '/'.join(reversed(
1212
[parent.name for parent in
1213
self._iter_file_id_parents(file_id)][:-1]))
1215
def path2id(self, name):
1216
"""Walk down through directories to return entry of last component.
1218
names may be either a list of path components, or a single
1219
string, in which case it is automatically split.
1221
This returns the entry of the last component in the path,
1222
which may be either a file or a directory.
1224
Returns None IFF the path is not found.
1226
if isinstance(name, basestring):
1227
name = osutils.splitpath(name)
1229
# mutter("lookup path %r" % name)
1236
children = getattr(parent, 'children', None)
1237
if children is None:
1240
assert cie.name == f
1241
assert cie.parent_id == parent.file_id
1244
# or raise an error?
1247
return parent.file_id
1249
def has_filename(self, names):
1250
return bool(self.path2id(names))
1252
def has_id(self, file_id):
1253
return (file_id in self._byid)
1255
def remove_recursive_id(self, file_id):
1256
"""Remove file_id, and children, from the inventory.
1258
:param file_id: A file_id to remove.
1260
to_find_delete = [self._byid[file_id]]
1262
while to_find_delete:
1263
ie = to_find_delete.pop()
1264
to_delete.append(ie.file_id)
1265
if ie.kind == 'directory':
1266
to_find_delete.extend(ie.children.values())
1267
for file_id in reversed(to_delete):
1269
del self._byid[file_id]
1270
if ie.parent_id is not None:
1271
del self[ie.parent_id].children[ie.name]
1273
def rename(self, file_id, new_parent_id, new_name):
1274
"""Move a file within the inventory.
1276
This can change either the name, or the parent, or both.
1278
This does not move the working file."""
1279
if not is_valid_name(new_name):
1280
raise BzrError("not an acceptable filename: %r" % new_name)
1282
new_parent = self._byid[new_parent_id]
1283
if new_name in new_parent.children:
1284
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
1286
new_parent_idpath = self.get_idpath(new_parent_id)
1287
if file_id in new_parent_idpath:
1288
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
1289
% (self.id2path(file_id), self.id2path(new_parent_id)))
1291
file_ie = self._byid[file_id]
1292
old_parent = self._byid[file_ie.parent_id]
1294
# TODO: Don't leave things messed up if this fails
1296
del old_parent.children[file_ie.name]
1297
new_parent.children[new_name] = file_ie
1299
file_ie.name = new_name
1300
file_ie.parent_id = new_parent_id
1302
def is_root(self, file_id):
1303
return self.root is not None and file_id == self.root.file_id
1306
def make_entry(kind, name, parent_id, file_id=None):
1307
"""Create an inventory entry.
1309
:param kind: the type of inventory entry to create.
1310
:param name: the basename of the entry.
1311
:param parent_id: the parent_id of the entry.
1312
:param file_id: the file_id to use. if None, one will be created.
1315
file_id = generate_ids.gen_file_id(name)
1317
norm_name, can_access = osutils.normalized_filename(name)
1318
if norm_name != name:
1322
# TODO: jam 20060701 This would probably be more useful
1323
# if the error was raised with the full path
1324
raise errors.InvalidNormalization(name)
1326
if kind == 'directory':
1327
return InventoryDirectory(file_id, name, parent_id)
1328
elif kind == 'file':
1329
return InventoryFile(file_id, name, parent_id)
1330
elif kind == 'symlink':
1331
return InventoryLink(file_id, name, parent_id)
1333
raise BzrError("unknown kind %r" % kind)
1338
def is_valid_name(name):
1340
if _NAME_RE is None:
1341
_NAME_RE = re.compile(r'^[^/\\]+$')
1343
return bool(_NAME_RE.match(name))
added
removed
bzrlib/inventory.py
