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(), """
47
from bzrlib.errors import (
51
from bzrlib.trace import mutter
54
class InventoryEntry(object):
55
"""Description of a versioned file.
57
An InventoryEntry has the following fields, which are also
58
present in the XML inventory-entry element:
63
(within the parent directory)
66
file_id of the parent directory, or ROOT_ID
69
the revision_id in which this variation of this file was
73
Indicates that this file should be executable on systems
77
sha-1 of the text of the file
80
size in bytes of the text of the file
82
(reading a version 4 tree created a text_id field.)
87
>>> i.add(InventoryDirectory('123', 'src', ROOT_ID))
88
InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None)
89
>>> i.add(InventoryFile('2323', 'hello.c', parent_id='123'))
90
InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None)
91
>>> shouldbe = {0: '', 1: 'src', 2: 'src/hello.c'}
92
>>> for ix, j in enumerate(i.iter_entries()):
93
... print (j[0] == shouldbe[ix], j[1])
95
(True, InventoryDirectory('TREE_ROOT', '', parent_id=None, revision=None))
96
(True, InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None))
97
(True, InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None))
98
>>> i.add(InventoryFile('2323', 'bye.c', '123'))
99
Traceback (most recent call last):
101
BzrError: inventory already contains entry with id {2323}
102
>>> i.add(InventoryFile('2324', 'bye.c', '123'))
103
InventoryFile('2324', 'bye.c', parent_id='123', sha1=None, len=None)
104
>>> i.add(InventoryDirectory('2325', 'wibble', '123'))
105
InventoryDirectory('2325', 'wibble', parent_id='123', revision=None)
106
>>> i.path2id('src/wibble')
110
>>> i.add(InventoryFile('2326', 'wibble.c', '2325'))
111
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
113
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
114
>>> for path, entry in i.iter_entries():
116
... assert i.path2id(path)
124
>>> i.id2path('2326')
125
'src/wibble/wibble.c'
128
# Constants returned by describe_change()
130
# TODO: These should probably move to some kind of FileChangeDescription
131
# class; that's like what's inside a TreeDelta but we want to be able to
132
# generate them just for one file at a time.
134
MODIFIED_AND_RENAMED = 'modified and renamed'
138
def detect_changes(self, old_entry):
139
"""Return a (text_modified, meta_modified) from this to old_entry.
141
_read_tree_state must have been called on self and old_entry prior to
142
calling detect_changes.
146
def diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
147
output_to, reverse=False):
148
"""Perform a diff from this to to_entry.
150
text_diff will be used for textual difference calculation.
151
This is a template method, override _diff in child classes.
153
self._read_tree_state(tree.id2path(self.file_id), tree)
155
# cannot diff from one kind to another - you must do a removal
156
# and an addif they do not match.
157
assert self.kind == to_entry.kind
158
to_entry._read_tree_state(to_tree.id2path(to_entry.file_id),
160
self._diff(text_diff, from_label, tree, to_label, to_entry, to_tree,
163
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
164
output_to, reverse=False):
165
"""Perform a diff between two entries of the same kind."""
167
def find_previous_heads(self, previous_inventories,
168
versioned_file_store,
171
"""Return the revisions and entries that directly precede this.
173
Returned as a map from revision to inventory entry.
175
This is a map containing the file revisions in all parents
176
for which the file exists, and its revision is not a parent of
177
any other. If the file is new, the set will be empty.
179
:param versioned_file_store: A store where ancestry data on this
180
file id can be queried.
181
:param transaction: The transaction that queries to the versioned
182
file store should be completed under.
183
:param entry_vf: The entry versioned file, if its already available.
185
def get_ancestors(weave, entry):
186
return set(weave.get_ancestry(entry.revision))
187
# revision:ie mapping for each ie found in previous_inventories.
189
# revision:ie mapping with one revision for each head.
191
# revision: ancestor list for each head
193
# identify candidate head revision ids.
194
for inv in previous_inventories:
195
if self.file_id in inv:
196
ie = inv[self.file_id]
197
assert ie.file_id == self.file_id
198
if ie.revision in candidates:
199
# same revision value in two different inventories:
200
# correct possible inconsistencies:
201
# * there was a bug in revision updates with 'x' bit
204
if candidates[ie.revision].executable != ie.executable:
205
candidates[ie.revision].executable = False
206
ie.executable = False
207
except AttributeError:
209
# must now be the same.
210
assert candidates[ie.revision] == ie
212
# add this revision as a candidate.
213
candidates[ie.revision] = ie
215
# common case optimisation
216
if len(candidates) == 1:
217
# if there is only one candidate revision found
218
# then we can opening the versioned file to access ancestry:
219
# there cannot be any ancestors to eliminate when there is
220
# only one revision available.
221
heads[ie.revision] = ie
224
# eliminate ancestors amongst the available candidates:
225
# heads are those that are not an ancestor of any other candidate
226
# - this provides convergence at a per-file level.
227
for ie in candidates.values():
228
# may be an ancestor of a known head:
229
already_present = 0 != len(
230
[head for head in heads
231
if ie.revision in head_ancestors[head]])
233
# an ancestor of an analyzed candidate.
235
# not an ancestor of a known head:
236
# load the versioned file for this file id if needed
238
entry_vf = versioned_file_store.get_weave_or_empty(
239
self.file_id, transaction)
240
ancestors = get_ancestors(entry_vf, ie)
241
# may knock something else out:
242
check_heads = list(heads.keys())
243
for head in check_heads:
244
if head in ancestors:
245
# this previously discovered 'head' is not
246
# really a head - its an ancestor of the newly
249
head_ancestors[ie.revision] = ancestors
250
heads[ie.revision] = ie
253
def get_tar_item(self, root, dp, now, tree):
254
"""Get a tarfile item and a file stream for its content."""
255
item = tarfile.TarInfo(osutils.pathjoin(root, dp).encode('utf8'))
256
# TODO: would be cool to actually set it to the timestamp of the
257
# revision it was last changed
259
fileobj = self._put_in_tar(item, tree)
263
"""Return true if the object this entry represents has textual data.
265
Note that textual data includes binary content.
267
Also note that all entries get weave files created for them.
268
This attribute is primarily used when upgrading from old trees that
269
did not have the weave index for all inventory entries.
273
def __init__(self, file_id, name, parent_id, text_id=None):
274
"""Create an InventoryEntry
276
The filename must be a single component, relative to the
277
parent directory; it cannot be a whole path or relative name.
279
>>> e = InventoryFile('123', 'hello.c', ROOT_ID)
284
>>> e = InventoryFile('123', 'src/hello.c', ROOT_ID)
285
Traceback (most recent call last):
286
InvalidEntryName: Invalid entry name: src/hello.c
288
assert isinstance(name, basestring), name
289
if '/' in name or '\\' in name:
290
raise errors.InvalidEntryName(name=name)
291
self.executable = False
293
self.text_sha1 = None
294
self.text_size = None
295
self.file_id = file_id
297
self.text_id = text_id
298
self.parent_id = parent_id
299
self.symlink_target = None
301
def kind_character(self):
302
"""Return a short kind indicator useful for appending to names."""
303
raise BzrError('unknown kind %r' % self.kind)
305
known_kinds = ('file', 'directory', 'symlink')
307
def _put_in_tar(self, item, tree):
308
"""populate item for stashing in a tar, and return the content stream.
310
If no content is available, return None.
312
raise BzrError("don't know how to export {%s} of kind %r" %
313
(self.file_id, self.kind))
315
def put_on_disk(self, dest, dp, tree):
316
"""Create a representation of self on disk in the prefix dest.
318
This is a template method - implement _put_on_disk in subclasses.
320
fullpath = osutils.pathjoin(dest, dp)
321
self._put_on_disk(fullpath, tree)
322
# mutter(" export {%s} kind %s to %s", self.file_id,
323
# self.kind, fullpath)
325
def _put_on_disk(self, fullpath, tree):
326
"""Put this entry onto disk at fullpath, from tree tree."""
327
raise BzrError("don't know how to export {%s} of kind %r" % (self.file_id, self.kind))
329
def sorted_children(self):
330
return sorted(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 is not 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, revision=%r)"
407
% (self.__class__.__name__,
413
def snapshot(self, revision, path, previous_entries,
414
work_tree, commit_builder):
415
"""Make a snapshot of this entry which may or may not have changed.
417
This means that all its fields are populated, that it has its
418
text stored in the text store or weave.
420
# mutter('new parents of %s are %r', path, previous_entries)
421
self._read_tree_state(path, work_tree)
422
# TODO: Where should we determine whether to reuse a
423
# previous revision id or create a new revision? 20060606
424
if len(previous_entries) == 1:
425
# cannot be unchanged unless there is only one parent file rev.
426
parent_ie = previous_entries.values()[0]
427
if self._unchanged(parent_ie):
428
# mutter("found unchanged entry")
429
self.revision = parent_ie.revision
431
return self._snapshot_into_revision(revision, previous_entries,
432
work_tree, commit_builder)
434
def _snapshot_into_revision(self, revision, previous_entries, work_tree,
436
"""Record this revision unconditionally into a store.
438
The entry's last-changed revision property (`revision`) is updated to
439
that of the new revision.
441
:param revision: id of the new revision that is being recorded.
443
:returns: String description of the commit (e.g. "merged", "modified"), etc.
445
# mutter('new revision {%s} for {%s}', revision, self.file_id)
446
self.revision = revision
447
self._snapshot_text(previous_entries, work_tree, commit_builder)
449
def _snapshot_text(self, file_parents, work_tree, commit_builder):
450
"""Record the 'text' of this entry, whatever form that takes.
452
This default implementation simply adds an empty text.
454
raise NotImplementedError(self._snapshot_text)
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 overridden 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 RootEntry(InventoryEntry):
509
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
510
'text_id', 'parent_id', 'children', 'executable',
511
'revision', 'symlink_target']
513
def _check(self, checker, rev_id, tree):
514
"""See InventoryEntry._check"""
516
def __init__(self, file_id):
517
self.file_id = file_id
519
self.kind = 'directory'
520
self.parent_id = None
523
symbol_versioning.warn('RootEntry is deprecated as of bzr 0.10.'
524
' Please use InventoryDirectory instead.',
525
DeprecationWarning, stacklevel=2)
527
def __eq__(self, other):
528
if not isinstance(other, RootEntry):
529
return NotImplemented
531
return (self.file_id == other.file_id) \
532
and (self.children == other.children)
535
class InventoryDirectory(InventoryEntry):
536
"""A directory in an inventory."""
538
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
539
'text_id', 'parent_id', 'children', 'executable',
540
'revision', 'symlink_target']
542
def _check(self, checker, rev_id, tree):
543
"""See InventoryEntry._check"""
544
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
545
raise BzrCheckError('directory {%s} has text in revision {%s}'
546
% (self.file_id, rev_id))
549
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
550
other.revision = self.revision
551
# note that children are *not* copied; they're pulled across when
555
def __init__(self, file_id, name, parent_id):
556
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
558
self.kind = 'directory'
560
def kind_character(self):
561
"""See InventoryEntry.kind_character."""
564
def _put_in_tar(self, item, tree):
565
"""See InventoryEntry._put_in_tar."""
566
item.type = tarfile.DIRTYPE
573
def _put_on_disk(self, fullpath, tree):
574
"""See InventoryEntry._put_on_disk."""
577
def _snapshot_text(self, file_parents, work_tree, commit_builder):
578
"""See InventoryEntry._snapshot_text."""
579
commit_builder.modified_directory(self.file_id, file_parents)
582
class InventoryFile(InventoryEntry):
583
"""A file in an inventory."""
585
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
586
'text_id', 'parent_id', 'children', 'executable',
587
'revision', 'symlink_target']
589
def _check(self, checker, tree_revision_id, tree):
590
"""See InventoryEntry._check"""
591
t = (self.file_id, self.revision)
592
if t in checker.checked_texts:
593
prev_sha = checker.checked_texts[t]
594
if prev_sha != self.text_sha1:
595
raise BzrCheckError('mismatched sha1 on {%s} in {%s}' %
596
(self.file_id, tree_revision_id))
598
checker.repeated_text_cnt += 1
601
if self.file_id not in checker.checked_weaves:
602
mutter('check weave {%s}', self.file_id)
603
w = tree.get_weave(self.file_id)
604
# Not passing a progress bar, because it creates a new
605
# progress, which overwrites the current progress,
606
# and doesn't look nice
608
checker.checked_weaves[self.file_id] = True
610
w = tree.get_weave(self.file_id)
612
mutter('check version {%s} of {%s}', tree_revision_id, self.file_id)
613
checker.checked_text_cnt += 1
614
# We can't check the length, because Weave doesn't store that
615
# information, and the whole point of looking at the weave's
616
# sha1sum is that we don't have to extract the text.
617
if self.text_sha1 != w.get_sha1(self.revision):
618
raise BzrCheckError('text {%s} version {%s} wrong sha1'
619
% (self.file_id, self.revision))
620
checker.checked_texts[t] = self.text_sha1
623
other = InventoryFile(self.file_id, self.name, self.parent_id)
624
other.executable = self.executable
625
other.text_id = self.text_id
626
other.text_sha1 = self.text_sha1
627
other.text_size = self.text_size
628
other.revision = self.revision
631
def detect_changes(self, old_entry):
632
"""See InventoryEntry.detect_changes."""
633
assert self.text_sha1 is not None
634
assert old_entry.text_sha1 is not None
635
text_modified = (self.text_sha1 != old_entry.text_sha1)
636
meta_modified = (self.executable != old_entry.executable)
637
return text_modified, meta_modified
639
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
640
output_to, reverse=False):
641
"""See InventoryEntry._diff."""
643
from_text = tree.get_file(self.file_id).readlines()
645
to_text = to_tree.get_file(to_entry.file_id).readlines()
649
text_diff(from_label, from_text,
650
to_label, to_text, output_to)
652
text_diff(to_label, to_text,
653
from_label, from_text, output_to)
654
except errors.BinaryFile:
656
label_pair = (to_label, from_label)
658
label_pair = (from_label, to_label)
659
print >> output_to, "Binary files %s and %s differ" % label_pair
662
"""See InventoryEntry.has_text."""
665
def __init__(self, file_id, name, parent_id):
666
super(InventoryFile, self).__init__(file_id, name, parent_id)
669
def kind_character(self):
670
"""See InventoryEntry.kind_character."""
673
def _put_in_tar(self, item, tree):
674
"""See InventoryEntry._put_in_tar."""
675
item.type = tarfile.REGTYPE
676
fileobj = tree.get_file(self.file_id)
677
item.size = self.text_size
678
if tree.is_executable(self.file_id):
684
def _put_on_disk(self, fullpath, tree):
685
"""See InventoryEntry._put_on_disk."""
686
osutils.pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
687
if tree.is_executable(self.file_id):
688
os.chmod(fullpath, 0755)
690
def _read_tree_state(self, path, work_tree):
691
"""See InventoryEntry._read_tree_state."""
692
self.text_sha1 = work_tree.get_file_sha1(self.file_id, path=path)
693
# FIXME: 20050930 probe for the text size when getting sha1
694
# in _read_tree_state
695
self.executable = work_tree.is_executable(self.file_id, path=path)
698
return ("%s(%r, %r, parent_id=%r, sha1=%r, len=%s)"
699
% (self.__class__.__name__,
706
def _forget_tree_state(self):
707
self.text_sha1 = None
709
def _snapshot_text(self, file_parents, work_tree, commit_builder):
710
"""See InventoryEntry._snapshot_text."""
711
def get_content_byte_lines():
712
return work_tree.get_file(self.file_id).readlines()
713
self.text_sha1, self.text_size = commit_builder.modified_file_text(
714
self.file_id, file_parents, get_content_byte_lines, self.text_sha1, self.text_size)
716
def _unchanged(self, previous_ie):
717
"""See InventoryEntry._unchanged."""
718
compatible = super(InventoryFile, self)._unchanged(previous_ie)
719
if self.text_sha1 != previous_ie.text_sha1:
722
# FIXME: 20050930 probe for the text size when getting sha1
723
# in _read_tree_state
724
self.text_size = previous_ie.text_size
725
if self.executable != previous_ie.executable:
730
class InventoryLink(InventoryEntry):
731
"""A file in an inventory."""
733
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
734
'text_id', 'parent_id', 'children', 'executable',
735
'revision', 'symlink_target']
737
def _check(self, checker, rev_id, tree):
738
"""See InventoryEntry._check"""
739
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
740
raise BzrCheckError('symlink {%s} has text in revision {%s}'
741
% (self.file_id, rev_id))
742
if self.symlink_target is None:
743
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
744
% (self.file_id, rev_id))
747
other = InventoryLink(self.file_id, self.name, self.parent_id)
748
other.symlink_target = self.symlink_target
749
other.revision = self.revision
752
def detect_changes(self, old_entry):
753
"""See InventoryEntry.detect_changes."""
754
# FIXME: which _modified field should we use ? RBC 20051003
755
text_modified = (self.symlink_target != old_entry.symlink_target)
757
mutter(" symlink target changed")
758
meta_modified = False
759
return text_modified, meta_modified
761
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
762
output_to, reverse=False):
763
"""See InventoryEntry._diff."""
764
from_text = self.symlink_target
765
if to_entry is not None:
766
to_text = to_entry.symlink_target
771
print >>output_to, '=== target changed %r => %r' % (from_text, to_text)
774
print >>output_to, '=== target was %r' % self.symlink_target
776
print >>output_to, '=== target is %r' % self.symlink_target
778
def __init__(self, file_id, name, parent_id):
779
super(InventoryLink, self).__init__(file_id, name, parent_id)
780
self.kind = 'symlink'
782
def kind_character(self):
783
"""See InventoryEntry.kind_character."""
786
def _put_in_tar(self, item, tree):
787
"""See InventoryEntry._put_in_tar."""
788
item.type = tarfile.SYMTYPE
792
item.linkname = self.symlink_target
795
def _put_on_disk(self, fullpath, tree):
796
"""See InventoryEntry._put_on_disk."""
798
os.symlink(self.symlink_target, fullpath)
800
raise BzrError("Failed to create symlink %r -> %r, error: %s" % (fullpath, self.symlink_target, e))
802
def _read_tree_state(self, path, work_tree):
803
"""See InventoryEntry._read_tree_state."""
804
self.symlink_target = work_tree.get_symlink_target(self.file_id)
806
def _forget_tree_state(self):
807
self.symlink_target = None
809
def _unchanged(self, previous_ie):
810
"""See InventoryEntry._unchanged."""
811
compatible = super(InventoryLink, self)._unchanged(previous_ie)
812
if self.symlink_target != previous_ie.symlink_target:
816
def _snapshot_text(self, file_parents, work_tree, commit_builder):
817
"""See InventoryEntry._snapshot_text."""
818
commit_builder.modified_link(
819
self.file_id, file_parents, self.symlink_target)
822
class Inventory(object):
823
"""Inventory of versioned files in a tree.
825
This describes which file_id is present at each point in the tree,
826
and possibly the SHA-1 or other information about the file.
827
Entries can be looked up either by path or by file_id.
829
The inventory represents a typical unix file tree, with
830
directories containing files and subdirectories. We never store
831
the full path to a file, because renaming a directory implicitly
832
moves all of its contents. This class internally maintains a
833
lookup tree that allows the children under a directory to be
836
InventoryEntry objects must not be modified after they are
837
inserted, other than through the Inventory API.
839
>>> inv = Inventory()
840
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
841
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
842
>>> inv['123-123'].name
845
May be treated as an iterator or set to look up file ids:
847
>>> bool(inv.path2id('hello.c'))
852
May also look up by name:
854
>>> [x[0] for x in inv.iter_entries()]
856
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
857
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
858
Traceback (most recent call last):
859
BzrError: parent_id {TREE_ROOT} not in inventory
860
>>> inv.add(InventoryFile('123-123', 'hello.c', 'TREE_ROOT-12345678-12345678'))
861
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT-12345678-12345678', sha1=None, len=None)
863
def __init__(self, root_id=ROOT_ID, revision_id=None):
864
"""Create or read an inventory.
866
If a working directory is specified, the inventory is read
867
from there. If the file is specified, read from that. If not,
868
the inventory is created empty.
870
The inventory is created with a default root directory, with
873
if root_id is not None:
874
self._set_root(InventoryDirectory(root_id, '', None))
878
self.revision_id = revision_id
880
def _set_root(self, ie):
882
self._byid = {self.root.file_id: self.root}
885
# TODO: jam 20051218 Should copy also copy the revision_id?
886
entries = self.iter_entries()
887
other = Inventory(entries.next()[1].file_id)
888
# copy recursively so we know directories will be added before
889
# their children. There are more efficient ways than this...
890
for path, entry in entries():
891
other.add(entry.copy())
895
return iter(self._byid)
898
"""Returns number of entries."""
899
return len(self._byid)
901
def iter_entries(self, from_dir=None):
902
"""Return (path, entry) pairs, in order by name."""
904
if self.root is None:
908
elif isinstance(from_dir, basestring):
909
from_dir = self._byid[from_dir]
911
# unrolling the recursive called changed the time from
912
# 440ms/663ms (inline/total) to 116ms/116ms
913
children = from_dir.children.items()
915
children = collections.deque(children)
916
stack = [(u'', children)]
918
from_dir_relpath, children = stack[-1]
921
name, ie = children.popleft()
923
# we know that from_dir_relpath never ends in a slash
924
# and 'f' doesn't begin with one, we can do a string op, rather
925
# than the checks of pathjoin(), though this means that all paths
927
path = from_dir_relpath + '/' + name
931
if ie.kind != 'directory':
934
# But do this child first
935
new_children = ie.children.items()
937
new_children = collections.deque(new_children)
938
stack.append((path, new_children))
939
# Break out of inner loop, so that we start outer loop with child
942
# if we finished all children, pop it off the stack
945
def iter_entries_by_dir(self, from_dir=None):
946
"""Iterate over the entries in a directory first order.
948
This returns all entries for a directory before returning
949
the entries for children of a directory. This is not
950
lexicographically sorted order, and is a hybrid between
951
depth-first and breadth-first.
953
:return: This yields (path, entry) pairs
955
# TODO? Perhaps this should return the from_dir so that the root is
956
# yielded? or maybe an option?
958
if self.root is None:
962
elif isinstance(from_dir, basestring):
963
from_dir = self._byid[from_dir]
965
stack = [(u'', from_dir)]
967
cur_relpath, cur_dir = stack.pop()
970
for child_name, child_ie in sorted(cur_dir.children.iteritems()):
972
child_relpath = cur_relpath + child_name
974
yield child_relpath, child_ie
976
if child_ie.kind == 'directory':
977
child_dirs.append((child_relpath+'/', child_ie))
978
stack.extend(reversed(child_dirs))
981
"""Return list of (path, ie) for all entries except the root.
983
This may be faster than iter_entries.
986
def descend(dir_ie, dir_path):
987
kids = dir_ie.children.items()
989
for name, ie in kids:
990
child_path = osutils.pathjoin(dir_path, name)
991
accum.append((child_path, ie))
992
if ie.kind == 'directory':
993
descend(ie, child_path)
995
descend(self.root, u'')
998
def directories(self):
999
"""Return (path, entry) pairs for all directories, including the root.
1002
def descend(parent_ie, parent_path):
1003
accum.append((parent_path, parent_ie))
1005
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
1008
for name, child_ie in kids:
1009
child_path = osutils.pathjoin(parent_path, name)
1010
descend(child_ie, child_path)
1011
descend(self.root, u'')
1014
def __contains__(self, file_id):
1015
"""True if this entry contains a file with given id.
1017
>>> inv = Inventory()
1018
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1019
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
1025
return (file_id in self._byid)
1027
def __getitem__(self, file_id):
1028
"""Return the entry for given file_id.
1030
>>> inv = Inventory()
1031
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
1032
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
1033
>>> inv['123123'].name
1037
return self._byid[file_id]
1040
raise BzrError("can't look up file_id None")
1042
raise BzrError("file_id {%s} not in inventory" % file_id)
1044
def get_file_kind(self, file_id):
1045
return self._byid[file_id].kind
1047
def get_child(self, parent_id, filename):
1048
return self[parent_id].children.get(filename)
1050
def add(self, entry):
1051
"""Add entry to inventory.
1053
To add a file to a branch ready to be committed, use Branch.add,
1056
Returns the new entry object.
1058
if entry.file_id in self._byid:
1059
raise BzrError("inventory already contains entry with id {%s}" % entry.file_id)
1061
if entry.parent_id is None:
1062
assert self.root is None and len(self._byid) == 0
1063
self._set_root(entry)
1066
parent = self._byid[entry.parent_id]
1068
raise BzrError("parent_id {%s} not in inventory" % entry.parent_id)
1070
if entry.name in parent.children:
1071
raise BzrError("%s is already versioned" %
1072
osutils.pathjoin(self.id2path(parent.file_id), entry.name))
1074
self._byid[entry.file_id] = entry
1075
parent.children[entry.name] = entry
1078
def add_path(self, relpath, kind, file_id=None, parent_id=None):
1079
"""Add entry from a path.
1081
The immediate parent must already be versioned.
1083
Returns the new entry object."""
1085
parts = osutils.splitpath(relpath)
1089
file_id = bzrlib.workingtree.gen_root_id()
1090
self.root = InventoryDirectory(file_id, '', None)
1091
self._byid = {self.root.file_id: self.root}
1094
parent_path = parts[:-1]
1095
parent_id = self.path2id(parent_path)
1096
if parent_id is None:
1097
raise errors.NotVersionedError(path=parent_path)
1098
ie = make_entry(kind, parts[-1], parent_id, file_id)
1101
def __delitem__(self, file_id):
1102
"""Remove entry by id.
1104
>>> inv = Inventory()
1105
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1106
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
1115
assert ie.parent_id is None or \
1116
self[ie.parent_id].children[ie.name] == ie
1118
del self._byid[file_id]
1119
if ie.parent_id is not None:
1120
del self[ie.parent_id].children[ie.name]
1122
def __eq__(self, other):
1123
"""Compare two sets by comparing their contents.
1125
>>> i1 = Inventory()
1126
>>> i2 = Inventory()
1129
>>> i1.add(InventoryFile('123', 'foo', ROOT_ID))
1130
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
1133
>>> i2.add(InventoryFile('123', 'foo', ROOT_ID))
1134
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
1138
if not isinstance(other, Inventory):
1139
return NotImplemented
1141
return self._byid == other._byid
1143
def __ne__(self, other):
1144
return not self.__eq__(other)
1147
raise ValueError('not hashable')
1149
def _iter_file_id_parents(self, file_id):
1150
"""Yield the parents of file_id up to the root."""
1151
while file_id is not None:
1153
ie = self._byid[file_id]
1155
raise BzrError("file_id {%s} not found in inventory" % file_id)
1157
file_id = ie.parent_id
1159
def get_idpath(self, file_id):
1160
"""Return a list of file_ids for the path to an entry.
1162
The list contains one element for each directory followed by
1163
the id of the file itself. So the length of the returned list
1164
is equal to the depth of the file in the tree, counting the
1165
root directory as depth 1.
1168
for parent in self._iter_file_id_parents(file_id):
1169
p.insert(0, parent.file_id)
1172
def id2path(self, file_id):
1173
"""Return as a string the path to file_id.
1176
>>> e = i.add(InventoryDirectory('src-id', 'src', ROOT_ID))
1177
>>> e = i.add(InventoryFile('foo-id', 'foo.c', parent_id='src-id'))
1178
>>> print i.id2path('foo-id')
1181
# get all names, skipping root
1182
return '/'.join(reversed(
1183
[parent.name for parent in
1184
self._iter_file_id_parents(file_id)][:-1]))
1186
def path2id(self, name):
1187
"""Walk down through directories to return entry of last component.
1189
names may be either a list of path components, or a single
1190
string, in which case it is automatically split.
1192
This returns the entry of the last component in the path,
1193
which may be either a file or a directory.
1195
Returns None IFF the path is not found.
1197
if isinstance(name, basestring):
1198
name = osutils.splitpath(name)
1200
# mutter("lookup path %r" % name)
1207
cie = parent.children[f]
1208
assert cie.name == f
1209
assert cie.parent_id == parent.file_id
1212
# or raise an error?
1215
return parent.file_id
1217
def has_filename(self, names):
1218
return bool(self.path2id(names))
1220
def has_id(self, file_id):
1221
return (file_id in self._byid)
1223
def remove_recursive_id(self, file_id):
1224
"""Remove file_id, and children, from the inventory.
1226
:param file_id: A file_id to remove.
1228
to_find_delete = [self._byid[file_id]]
1230
while to_find_delete:
1231
ie = to_find_delete.pop()
1232
to_delete.append(ie.file_id)
1233
if ie.kind == 'directory':
1234
to_find_delete.extend(ie.children.values())
1235
for file_id in reversed(to_delete):
1237
del self._byid[file_id]
1238
if ie.parent_id is not None:
1239
del self[ie.parent_id].children[ie.name]
1241
def rename(self, file_id, new_parent_id, new_name):
1242
"""Move a file within the inventory.
1244
This can change either the name, or the parent, or both.
1246
This does not move the working file."""
1247
if not is_valid_name(new_name):
1248
raise BzrError("not an acceptable filename: %r" % new_name)
1250
new_parent = self._byid[new_parent_id]
1251
if new_name in new_parent.children:
1252
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
1254
new_parent_idpath = self.get_idpath(new_parent_id)
1255
if file_id in new_parent_idpath:
1256
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
1257
% (self.id2path(file_id), self.id2path(new_parent_id)))
1259
file_ie = self._byid[file_id]
1260
old_parent = self._byid[file_ie.parent_id]
1262
# TODO: Don't leave things messed up if this fails
1264
del old_parent.children[file_ie.name]
1265
new_parent.children[new_name] = file_ie
1267
file_ie.name = new_name
1268
file_ie.parent_id = new_parent_id
1270
def is_root(self, file_id):
1271
return self.root is not None and file_id == self.root.file_id
1274
def make_entry(kind, name, parent_id, file_id=None):
1275
"""Create an inventory entry.
1277
:param kind: the type of inventory entry to create.
1278
:param name: the basename of the entry.
1279
:param parent_id: the parent_id of the entry.
1280
:param file_id: the file_id to use. if None, one will be created.
1283
file_id = bzrlib.workingtree.gen_file_id(name)
1285
norm_name, can_access = osutils.normalized_filename(name)
1286
if norm_name != name:
1290
# TODO: jam 20060701 This would probably be more useful
1291
# if the error was raised with the full path
1292
raise errors.InvalidNormalization(name)
1294
if kind == 'directory':
1295
return InventoryDirectory(file_id, name, parent_id)
1296
elif kind == 'file':
1297
return InventoryFile(file_id, name, parent_id)
1298
elif kind == 'symlink':
1299
return InventoryLink(file_id, name, parent_id)
1301
raise BzrError("unknown kind %r" % kind)
1306
def is_valid_name(name):
1308
if _NAME_RE is None:
1309
_NAME_RE = re.compile(r'^[^/\\]+$')
1311
return bool(_NAME_RE.match(name))
added
removed
bzrlib/inventory.py
