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(), """
48
from bzrlib.errors import (
52
from bzrlib.trace import mutter
55
class InventoryEntry(object):
56
"""Description of a versioned file.
58
An InventoryEntry has the following fields, which are also
59
present in the XML inventory-entry element:
64
(within the parent directory)
67
file_id of the parent directory, or ROOT_ID
70
the revision_id in which this variation of this file was
74
Indicates that this file should be executable on systems
78
sha-1 of the text of the file
81
size in bytes of the text of the file
83
(reading a version 4 tree created a text_id field.)
88
>>> i.add(InventoryDirectory('123', 'src', ROOT_ID))
89
InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None)
90
>>> i.add(InventoryFile('2323', 'hello.c', parent_id='123'))
91
InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None)
92
>>> shouldbe = {0: '', 1: 'src', 2: 'src/hello.c'}
93
>>> for ix, j in enumerate(i.iter_entries()):
94
... print (j[0] == shouldbe[ix], j[1])
96
(True, InventoryDirectory('TREE_ROOT', '', parent_id=None, revision=None))
97
(True, InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None))
98
(True, InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None))
99
>>> i.add(InventoryFile('2323', 'bye.c', '123'))
100
Traceback (most recent call last):
102
BzrError: inventory already contains entry with id {2323}
103
>>> i.add(InventoryFile('2324', 'bye.c', '123'))
104
InventoryFile('2324', 'bye.c', parent_id='123', sha1=None, len=None)
105
>>> i.add(InventoryDirectory('2325', 'wibble', '123'))
106
InventoryDirectory('2325', 'wibble', parent_id='123', revision=None)
107
>>> i.path2id('src/wibble')
111
>>> i.add(InventoryFile('2326', 'wibble.c', '2325'))
112
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
114
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
115
>>> for path, entry in i.iter_entries():
117
... assert i.path2id(path)
125
>>> i.id2path('2326')
126
'src/wibble/wibble.c'
129
# Constants returned by describe_change()
131
# TODO: These should probably move to some kind of FileChangeDescription
132
# class; that's like what's inside a TreeDelta but we want to be able to
133
# generate them just for one file at a time.
135
MODIFIED_AND_RENAMED = 'modified and renamed'
139
def detect_changes(self, old_entry):
140
"""Return a (text_modified, meta_modified) from this to old_entry.
142
_read_tree_state must have been called on self and old_entry prior to
143
calling detect_changes.
147
def diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
148
output_to, reverse=False):
149
"""Perform a diff from this to to_entry.
151
text_diff will be used for textual difference calculation.
152
This is a template method, override _diff in child classes.
154
self._read_tree_state(tree.id2path(self.file_id), tree)
156
# cannot diff from one kind to another - you must do a removal
157
# and an addif they do not match.
158
assert self.kind == to_entry.kind
159
to_entry._read_tree_state(to_tree.id2path(to_entry.file_id),
161
self._diff(text_diff, from_label, tree, to_label, to_entry, to_tree,
164
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
165
output_to, reverse=False):
166
"""Perform a diff between two entries of the same kind."""
168
def find_previous_heads(self, previous_inventories,
169
versioned_file_store,
172
"""Return the revisions and entries that directly precede this.
174
Returned as a map from revision to inventory entry.
176
This is a map containing the file revisions in all parents
177
for which the file exists, and its revision is not a parent of
178
any other. If the file is new, the set will be empty.
180
:param versioned_file_store: A store where ancestry data on this
181
file id can be queried.
182
:param transaction: The transaction that queries to the versioned
183
file store should be completed under.
184
:param entry_vf: The entry versioned file, if its already available.
186
def get_ancestors(weave, entry):
187
return set(weave.get_ancestry(entry.revision))
188
# revision:ie mapping for each ie found in previous_inventories.
190
# revision:ie mapping with one revision for each head.
192
# revision: ancestor list for each head
194
# identify candidate head revision ids.
195
for inv in previous_inventories:
196
if self.file_id in inv:
197
ie = inv[self.file_id]
198
assert ie.file_id == self.file_id
199
if ie.revision in candidates:
200
# same revision value in two different inventories:
201
# correct possible inconsistencies:
202
# * there was a bug in revision updates with 'x' bit
205
if candidates[ie.revision].executable != ie.executable:
206
candidates[ie.revision].executable = False
207
ie.executable = False
208
except AttributeError:
210
# must now be the same.
211
assert candidates[ie.revision] == ie
213
# add this revision as a candidate.
214
candidates[ie.revision] = ie
216
# common case optimisation
217
if len(candidates) == 1:
218
# if there is only one candidate revision found
219
# then we can opening the versioned file to access ancestry:
220
# there cannot be any ancestors to eliminate when there is
221
# only one revision available.
222
heads[ie.revision] = ie
225
# eliminate ancestors amongst the available candidates:
226
# heads are those that are not an ancestor of any other candidate
227
# - this provides convergence at a per-file level.
228
for ie in candidates.values():
229
# may be an ancestor of a known head:
230
already_present = 0 != len(
231
[head for head in heads
232
if ie.revision in head_ancestors[head]])
234
# an ancestor of an analyzed candidate.
236
# not an ancestor of a known head:
237
# load the versioned file for this file id if needed
239
entry_vf = versioned_file_store.get_weave_or_empty(
240
self.file_id, transaction)
241
ancestors = get_ancestors(entry_vf, ie)
242
# may knock something else out:
243
check_heads = list(heads.keys())
244
for head in check_heads:
245
if head in ancestors:
246
# this previously discovered 'head' is not
247
# really a head - its an ancestor of the newly
250
head_ancestors[ie.revision] = ancestors
251
heads[ie.revision] = ie
254
def get_tar_item(self, root, dp, now, tree):
255
"""Get a tarfile item and a file stream for its content."""
256
item = tarfile.TarInfo(osutils.pathjoin(root, dp).encode('utf8'))
257
# TODO: would be cool to actually set it to the timestamp of the
258
# revision it was last changed
260
fileobj = self._put_in_tar(item, tree)
264
"""Return true if the object this entry represents has textual data.
266
Note that textual data includes binary content.
268
Also note that all entries get weave files created for them.
269
This attribute is primarily used when upgrading from old trees that
270
did not have the weave index for all inventory entries.
274
def __init__(self, file_id, name, parent_id, text_id=None):
275
"""Create an InventoryEntry
277
The filename must be a single component, relative to the
278
parent directory; it cannot be a whole path or relative name.
280
>>> e = InventoryFile('123', 'hello.c', ROOT_ID)
285
>>> e = InventoryFile('123', 'src/hello.c', ROOT_ID)
286
Traceback (most recent call last):
287
InvalidEntryName: Invalid entry name: src/hello.c
289
assert isinstance(name, basestring), name
290
if '/' in name or '\\' in name:
291
raise errors.InvalidEntryName(name=name)
292
self.executable = False
294
self.text_sha1 = None
295
self.text_size = None
296
self.file_id = file_id
298
self.text_id = text_id
299
self.parent_id = parent_id
300
self.symlink_target = None
302
def kind_character(self):
303
"""Return a short kind indicator useful for appending to names."""
304
raise BzrError('unknown kind %r' % self.kind)
306
known_kinds = ('file', 'directory', 'symlink')
308
def _put_in_tar(self, item, tree):
309
"""populate item for stashing in a tar, and return the content stream.
311
If no content is available, return None.
313
raise BzrError("don't know how to export {%s} of kind %r" %
314
(self.file_id, self.kind))
316
def put_on_disk(self, dest, dp, tree):
317
"""Create a representation of self on disk in the prefix dest.
319
This is a template method - implement _put_on_disk in subclasses.
321
fullpath = osutils.pathjoin(dest, dp)
322
self._put_on_disk(fullpath, tree)
323
# mutter(" export {%s} kind %s to %s", self.file_id,
324
# self.kind, fullpath)
326
def _put_on_disk(self, fullpath, tree):
327
"""Put this entry onto disk at fullpath, from tree tree."""
328
raise BzrError("don't know how to export {%s} of kind %r" % (self.file_id, self.kind))
330
def sorted_children(self):
331
return sorted(self.children.items())
334
def versionable_kind(kind):
335
return (kind in ('file', 'directory', 'symlink'))
337
def check(self, checker, rev_id, inv, tree):
338
"""Check this inventory entry is intact.
340
This is a template method, override _check for kind specific
343
:param checker: Check object providing context for the checks;
344
can be used to find out what parts of the repository have already
346
:param rev_id: Revision id from which this InventoryEntry was loaded.
347
Not necessarily the last-changed revision for this file.
348
:param inv: Inventory from which the entry was loaded.
349
:param tree: RevisionTree for this entry.
351
if self.parent_id is not None:
352
if not inv.has_id(self.parent_id):
353
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
354
% (self.parent_id, rev_id))
355
self._check(checker, rev_id, tree)
357
def _check(self, checker, rev_id, tree):
358
"""Check this inventory entry for kind specific errors."""
359
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
363
"""Clone this inventory entry."""
364
raise NotImplementedError
367
def describe_change(old_entry, new_entry):
368
"""Describe the change between old_entry and this.
370
This smells of being an InterInventoryEntry situation, but as its
371
the first one, we're making it a static method for now.
373
An entry with a different parent, or different name is considered
374
to be renamed. Reparenting is an internal detail.
375
Note that renaming the parent does not trigger a rename for the
378
# TODO: Perhaps return an object rather than just a string
379
if old_entry is new_entry:
380
# also the case of both being None
382
elif old_entry is None:
384
elif new_entry is None:
386
text_modified, meta_modified = new_entry.detect_changes(old_entry)
387
if text_modified or meta_modified:
391
# TODO 20060511 (mbp, rbc) factor out 'detect_rename' here.
392
if old_entry.parent_id != new_entry.parent_id:
394
elif old_entry.name != new_entry.name:
398
if renamed and not modified:
399
return InventoryEntry.RENAMED
400
if modified and not renamed:
402
if modified and renamed:
403
return InventoryEntry.MODIFIED_AND_RENAMED
407
return ("%s(%r, %r, parent_id=%r, revision=%r)"
408
% (self.__class__.__name__,
414
def snapshot(self, revision, path, previous_entries,
415
work_tree, commit_builder):
416
"""Make a snapshot of this entry which may or may not have changed.
418
This means that all its fields are populated, that it has its
419
text stored in the text store or weave.
421
# mutter('new parents of %s are %r', path, previous_entries)
422
self._read_tree_state(path, work_tree)
423
# TODO: Where should we determine whether to reuse a
424
# previous revision id or create a new revision? 20060606
425
if len(previous_entries) == 1:
426
# cannot be unchanged unless there is only one parent file rev.
427
parent_ie = previous_entries.values()[0]
428
if self._unchanged(parent_ie):
429
# mutter("found unchanged entry")
430
self.revision = parent_ie.revision
432
return self._snapshot_into_revision(revision, previous_entries,
433
work_tree, commit_builder)
435
def _snapshot_into_revision(self, revision, previous_entries, work_tree,
437
"""Record this revision unconditionally into a store.
439
The entry's last-changed revision property (`revision`) is updated to
440
that of the new revision.
442
:param revision: id of the new revision that is being recorded.
444
:returns: String description of the commit (e.g. "merged", "modified"), etc.
446
# mutter('new revision {%s} for {%s}', revision, self.file_id)
447
self.revision = revision
448
self._snapshot_text(previous_entries, work_tree, commit_builder)
450
def _snapshot_text(self, file_parents, work_tree, commit_builder):
451
"""Record the 'text' of this entry, whatever form that takes.
453
This default implementation simply adds an empty text.
455
raise NotImplementedError(self._snapshot_text)
457
def __eq__(self, other):
458
if not isinstance(other, InventoryEntry):
459
return NotImplemented
461
return ((self.file_id == other.file_id)
462
and (self.name == other.name)
463
and (other.symlink_target == self.symlink_target)
464
and (self.text_sha1 == other.text_sha1)
465
and (self.text_size == other.text_size)
466
and (self.text_id == other.text_id)
467
and (self.parent_id == other.parent_id)
468
and (self.kind == other.kind)
469
and (self.revision == other.revision)
470
and (self.executable == other.executable)
473
def __ne__(self, other):
474
return not (self == other)
477
raise ValueError('not hashable')
479
def _unchanged(self, previous_ie):
480
"""Has this entry changed relative to previous_ie.
482
This method should be overridden in child classes.
485
# different inv parent
486
if previous_ie.parent_id != self.parent_id:
489
elif previous_ie.name != self.name:
493
def _read_tree_state(self, path, work_tree):
494
"""Populate fields in the inventory entry from the given tree.
496
Note that this should be modified to be a noop on virtual trees
497
as all entries created there are prepopulated.
499
# TODO: Rather than running this manually, we should check the
500
# working sha1 and other expensive properties when they're
501
# first requested, or preload them if they're already known
502
pass # nothing to do by default
504
def _forget_tree_state(self):
508
class RootEntry(InventoryEntry):
510
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
511
'text_id', 'parent_id', 'children', 'executable',
512
'revision', 'symlink_target']
514
def _check(self, checker, rev_id, tree):
515
"""See InventoryEntry._check"""
517
def __init__(self, file_id):
518
self.file_id = file_id
520
self.kind = 'directory'
521
self.parent_id = None
524
symbol_versioning.warn('RootEntry is deprecated as of bzr 0.10.'
525
' Please use InventoryDirectory instead.',
526
DeprecationWarning, stacklevel=2)
528
def __eq__(self, other):
529
if not isinstance(other, RootEntry):
530
return NotImplemented
532
return (self.file_id == other.file_id) \
533
and (self.children == other.children)
536
class InventoryDirectory(InventoryEntry):
537
"""A directory in an inventory."""
539
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
540
'text_id', 'parent_id', 'children', 'executable',
541
'revision', 'symlink_target']
543
def _check(self, checker, rev_id, tree):
544
"""See InventoryEntry._check"""
545
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
546
raise BzrCheckError('directory {%s} has text in revision {%s}'
547
% (self.file_id, rev_id))
550
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
551
other.revision = self.revision
552
# note that children are *not* copied; they're pulled across when
556
def __init__(self, file_id, name, parent_id):
557
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
559
self.kind = 'directory'
561
def kind_character(self):
562
"""See InventoryEntry.kind_character."""
565
def _put_in_tar(self, item, tree):
566
"""See InventoryEntry._put_in_tar."""
567
item.type = tarfile.DIRTYPE
574
def _put_on_disk(self, fullpath, tree):
575
"""See InventoryEntry._put_on_disk."""
578
def _snapshot_text(self, file_parents, work_tree, commit_builder):
579
"""See InventoryEntry._snapshot_text."""
580
commit_builder.modified_directory(self.file_id, file_parents)
583
class InventoryFile(InventoryEntry):
584
"""A file in an inventory."""
586
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
587
'text_id', 'parent_id', 'children', 'executable',
588
'revision', 'symlink_target']
590
def _check(self, checker, tree_revision_id, tree):
591
"""See InventoryEntry._check"""
592
t = (self.file_id, self.revision)
593
if t in checker.checked_texts:
594
prev_sha = checker.checked_texts[t]
595
if prev_sha != self.text_sha1:
596
raise BzrCheckError('mismatched sha1 on {%s} in {%s}' %
597
(self.file_id, tree_revision_id))
599
checker.repeated_text_cnt += 1
602
if self.file_id not in checker.checked_weaves:
603
mutter('check weave {%s}', self.file_id)
604
w = tree.get_weave(self.file_id)
605
# Not passing a progress bar, because it creates a new
606
# progress, which overwrites the current progress,
607
# and doesn't look nice
609
checker.checked_weaves[self.file_id] = True
611
w = tree.get_weave(self.file_id)
613
mutter('check version {%s} of {%s}', tree_revision_id, self.file_id)
614
checker.checked_text_cnt += 1
615
# We can't check the length, because Weave doesn't store that
616
# information, and the whole point of looking at the weave's
617
# sha1sum is that we don't have to extract the text.
618
if self.text_sha1 != w.get_sha1(self.revision):
619
raise BzrCheckError('text {%s} version {%s} wrong sha1'
620
% (self.file_id, self.revision))
621
checker.checked_texts[t] = self.text_sha1
624
other = InventoryFile(self.file_id, self.name, self.parent_id)
625
other.executable = self.executable
626
other.text_id = self.text_id
627
other.text_sha1 = self.text_sha1
628
other.text_size = self.text_size
629
other.revision = self.revision
632
def detect_changes(self, old_entry):
633
"""See InventoryEntry.detect_changes."""
634
assert self.text_sha1 is not None
635
assert old_entry.text_sha1 is not None
636
text_modified = (self.text_sha1 != old_entry.text_sha1)
637
meta_modified = (self.executable != old_entry.executable)
638
return text_modified, meta_modified
640
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
641
output_to, reverse=False):
642
"""See InventoryEntry._diff."""
644
from_text = tree.get_file(self.file_id).readlines()
646
to_text = to_tree.get_file(to_entry.file_id).readlines()
650
text_diff(from_label, from_text,
651
to_label, to_text, output_to)
653
text_diff(to_label, to_text,
654
from_label, from_text, output_to)
655
except errors.BinaryFile:
657
label_pair = (to_label, from_label)
659
label_pair = (from_label, to_label)
660
print >> output_to, "Binary files %s and %s differ" % label_pair
663
"""See InventoryEntry.has_text."""
666
def __init__(self, file_id, name, parent_id):
667
super(InventoryFile, self).__init__(file_id, name, parent_id)
670
def kind_character(self):
671
"""See InventoryEntry.kind_character."""
674
def _put_in_tar(self, item, tree):
675
"""See InventoryEntry._put_in_tar."""
676
item.type = tarfile.REGTYPE
677
fileobj = tree.get_file(self.file_id)
678
item.size = self.text_size
679
if tree.is_executable(self.file_id):
685
def _put_on_disk(self, fullpath, tree):
686
"""See InventoryEntry._put_on_disk."""
687
osutils.pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
688
if tree.is_executable(self.file_id):
689
os.chmod(fullpath, 0755)
691
def _read_tree_state(self, path, work_tree):
692
"""See InventoryEntry._read_tree_state."""
693
self.text_sha1 = work_tree.get_file_sha1(self.file_id, path=path)
694
# FIXME: 20050930 probe for the text size when getting sha1
695
# in _read_tree_state
696
self.executable = work_tree.is_executable(self.file_id, path=path)
699
return ("%s(%r, %r, parent_id=%r, sha1=%r, len=%s)"
700
% (self.__class__.__name__,
707
def _forget_tree_state(self):
708
self.text_sha1 = None
710
def _snapshot_text(self, file_parents, work_tree, commit_builder):
711
"""See InventoryEntry._snapshot_text."""
712
def get_content_byte_lines():
713
return work_tree.get_file(self.file_id).readlines()
714
self.text_sha1, self.text_size = commit_builder.modified_file_text(
715
self.file_id, file_parents, get_content_byte_lines, self.text_sha1, self.text_size)
717
def _unchanged(self, previous_ie):
718
"""See InventoryEntry._unchanged."""
719
compatible = super(InventoryFile, self)._unchanged(previous_ie)
720
if self.text_sha1 != previous_ie.text_sha1:
723
# FIXME: 20050930 probe for the text size when getting sha1
724
# in _read_tree_state
725
self.text_size = previous_ie.text_size
726
if self.executable != previous_ie.executable:
731
class InventoryLink(InventoryEntry):
732
"""A file in an inventory."""
734
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
735
'text_id', 'parent_id', 'children', 'executable',
736
'revision', 'symlink_target']
738
def _check(self, checker, rev_id, tree):
739
"""See InventoryEntry._check"""
740
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
741
raise BzrCheckError('symlink {%s} has text in revision {%s}'
742
% (self.file_id, rev_id))
743
if self.symlink_target is None:
744
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
745
% (self.file_id, rev_id))
748
other = InventoryLink(self.file_id, self.name, self.parent_id)
749
other.symlink_target = self.symlink_target
750
other.revision = self.revision
753
def detect_changes(self, old_entry):
754
"""See InventoryEntry.detect_changes."""
755
# FIXME: which _modified field should we use ? RBC 20051003
756
text_modified = (self.symlink_target != old_entry.symlink_target)
758
mutter(" symlink target changed")
759
meta_modified = False
760
return text_modified, meta_modified
762
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
763
output_to, reverse=False):
764
"""See InventoryEntry._diff."""
765
from_text = self.symlink_target
766
if to_entry is not None:
767
to_text = to_entry.symlink_target
772
print >>output_to, '=== target changed %r => %r' % (from_text, to_text)
775
print >>output_to, '=== target was %r' % self.symlink_target
777
print >>output_to, '=== target is %r' % self.symlink_target
779
def __init__(self, file_id, name, parent_id):
780
super(InventoryLink, self).__init__(file_id, name, parent_id)
781
self.kind = 'symlink'
783
def kind_character(self):
784
"""See InventoryEntry.kind_character."""
787
def _put_in_tar(self, item, tree):
788
"""See InventoryEntry._put_in_tar."""
789
item.type = tarfile.SYMTYPE
793
item.linkname = self.symlink_target
796
def _put_on_disk(self, fullpath, tree):
797
"""See InventoryEntry._put_on_disk."""
799
os.symlink(self.symlink_target, fullpath)
801
raise BzrError("Failed to create symlink %r -> %r, error: %s" % (fullpath, self.symlink_target, e))
803
def _read_tree_state(self, path, work_tree):
804
"""See InventoryEntry._read_tree_state."""
805
self.symlink_target = work_tree.get_symlink_target(self.file_id)
807
def _forget_tree_state(self):
808
self.symlink_target = None
810
def _unchanged(self, previous_ie):
811
"""See InventoryEntry._unchanged."""
812
compatible = super(InventoryLink, self)._unchanged(previous_ie)
813
if self.symlink_target != previous_ie.symlink_target:
817
def _snapshot_text(self, file_parents, work_tree, commit_builder):
818
"""See InventoryEntry._snapshot_text."""
819
commit_builder.modified_link(
820
self.file_id, file_parents, self.symlink_target)
823
class Inventory(object):
824
"""Inventory of versioned files in a tree.
826
This describes which file_id is present at each point in the tree,
827
and possibly the SHA-1 or other information about the file.
828
Entries can be looked up either by path or by file_id.
830
The inventory represents a typical unix file tree, with
831
directories containing files and subdirectories. We never store
832
the full path to a file, because renaming a directory implicitly
833
moves all of its contents. This class internally maintains a
834
lookup tree that allows the children under a directory to be
837
InventoryEntry objects must not be modified after they are
838
inserted, other than through the Inventory API.
840
>>> inv = Inventory()
841
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
842
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
843
>>> inv['123-123'].name
846
May be treated as an iterator or set to look up file ids:
848
>>> bool(inv.path2id('hello.c'))
853
May also look up by name:
855
>>> [x[0] for x in inv.iter_entries()]
857
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
858
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
859
Traceback (most recent call last):
860
BzrError: parent_id {TREE_ROOT} not in inventory
861
>>> inv.add(InventoryFile('123-123', 'hello.c', 'TREE_ROOT-12345678-12345678'))
862
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT-12345678-12345678', sha1=None, len=None)
864
def __init__(self, root_id=ROOT_ID, revision_id=None):
865
"""Create or read an inventory.
867
If a working directory is specified, the inventory is read
868
from there. If the file is specified, read from that. If not,
869
the inventory is created empty.
871
The inventory is created with a default root directory, with
874
if root_id is not None:
875
self._set_root(InventoryDirectory(root_id, '', None))
879
self.revision_id = revision_id
881
def _set_root(self, ie):
883
self._byid = {self.root.file_id: self.root}
886
# TODO: jam 20051218 Should copy also copy the revision_id?
887
entries = self.iter_entries()
888
other = Inventory(entries.next()[1].file_id)
889
# copy recursively so we know directories will be added before
890
# their children. There are more efficient ways than this...
891
for path, entry in entries():
892
other.add(entry.copy())
896
return iter(self._byid)
899
"""Returns number of entries."""
900
return len(self._byid)
902
def iter_entries(self, from_dir=None):
903
"""Return (path, entry) pairs, in order by name."""
905
if self.root is None:
909
elif isinstance(from_dir, basestring):
910
from_dir = self._byid[from_dir]
912
# unrolling the recursive called changed the time from
913
# 440ms/663ms (inline/total) to 116ms/116ms
914
children = from_dir.children.items()
916
children = collections.deque(children)
917
stack = [(u'', children)]
919
from_dir_relpath, children = stack[-1]
922
name, ie = children.popleft()
924
# we know that from_dir_relpath never ends in a slash
925
# and 'f' doesn't begin with one, we can do a string op, rather
926
# than the checks of pathjoin(), though this means that all paths
928
path = from_dir_relpath + '/' + name
932
if ie.kind != 'directory':
935
# But do this child first
936
new_children = ie.children.items()
938
new_children = collections.deque(new_children)
939
stack.append((path, new_children))
940
# Break out of inner loop, so that we start outer loop with child
943
# if we finished all children, pop it off the stack
946
def iter_entries_by_dir(self, from_dir=None):
947
"""Iterate over the entries in a directory first order.
949
This returns all entries for a directory before returning
950
the entries for children of a directory. This is not
951
lexicographically sorted order, and is a hybrid between
952
depth-first and breadth-first.
954
:return: This yields (path, entry) pairs
956
# TODO? Perhaps this should return the from_dir so that the root is
957
# yielded? or maybe an option?
959
if self.root is None:
963
elif isinstance(from_dir, basestring):
964
from_dir = self._byid[from_dir]
966
stack = [(u'', from_dir)]
968
cur_relpath, cur_dir = stack.pop()
971
for child_name, child_ie in sorted(cur_dir.children.iteritems()):
973
child_relpath = cur_relpath + child_name
975
yield child_relpath, child_ie
977
if child_ie.kind == 'directory':
978
child_dirs.append((child_relpath+'/', child_ie))
979
stack.extend(reversed(child_dirs))
982
"""Return list of (path, ie) for all entries except the root.
984
This may be faster than iter_entries.
987
def descend(dir_ie, dir_path):
988
kids = dir_ie.children.items()
990
for name, ie in kids:
991
child_path = osutils.pathjoin(dir_path, name)
992
accum.append((child_path, ie))
993
if ie.kind == 'directory':
994
descend(ie, child_path)
996
descend(self.root, u'')
999
def directories(self):
1000
"""Return (path, entry) pairs for all directories, including the root.
1003
def descend(parent_ie, parent_path):
1004
accum.append((parent_path, parent_ie))
1006
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
1009
for name, child_ie in kids:
1010
child_path = osutils.pathjoin(parent_path, name)
1011
descend(child_ie, child_path)
1012
descend(self.root, u'')
1015
def __contains__(self, file_id):
1016
"""True if this entry contains a file with given id.
1018
>>> inv = Inventory()
1019
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1020
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
1026
return (file_id in self._byid)
1028
def __getitem__(self, file_id):
1029
"""Return the entry for given file_id.
1031
>>> inv = Inventory()
1032
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
1033
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
1034
>>> inv['123123'].name
1038
return self._byid[file_id]
1041
raise BzrError("can't look up file_id None")
1043
raise BzrError("file_id {%s} not in inventory" % file_id)
1045
def get_file_kind(self, file_id):
1046
return self._byid[file_id].kind
1048
def get_child(self, parent_id, filename):
1049
return self[parent_id].children.get(filename)
1051
def add(self, entry):
1052
"""Add entry to inventory.
1054
To add a file to a branch ready to be committed, use Branch.add,
1057
Returns the new entry object.
1059
if entry.file_id in self._byid:
1060
raise BzrError("inventory already contains entry with id {%s}" % entry.file_id)
1062
if entry.parent_id is None:
1063
assert self.root is None and len(self._byid) == 0
1064
self._set_root(entry)
1067
parent = self._byid[entry.parent_id]
1069
raise BzrError("parent_id {%s} not in inventory" % entry.parent_id)
1071
if entry.name in parent.children:
1072
raise BzrError("%s is already versioned" %
1073
osutils.pathjoin(self.id2path(parent.file_id), entry.name))
1075
self._byid[entry.file_id] = entry
1076
parent.children[entry.name] = entry
1079
def add_path(self, relpath, kind, file_id=None, parent_id=None):
1080
"""Add entry from a path.
1082
The immediate parent must already be versioned.
1084
Returns the new entry object."""
1086
parts = osutils.splitpath(relpath)
1090
file_id = workingtree.gen_root_id()
1091
self.root = InventoryDirectory(file_id, '', None)
1092
self._byid = {self.root.file_id: self.root}
1095
parent_path = parts[:-1]
1096
parent_id = self.path2id(parent_path)
1097
if parent_id is None:
1098
raise errors.NotVersionedError(path=parent_path)
1099
ie = make_entry(kind, parts[-1], parent_id, file_id)
1102
def __delitem__(self, file_id):
1103
"""Remove entry by id.
1105
>>> inv = Inventory()
1106
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1107
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
1116
assert ie.parent_id is None or \
1117
self[ie.parent_id].children[ie.name] == ie
1119
del self._byid[file_id]
1120
if ie.parent_id is not None:
1121
del self[ie.parent_id].children[ie.name]
1123
def __eq__(self, other):
1124
"""Compare two sets by comparing their contents.
1126
>>> i1 = Inventory()
1127
>>> i2 = Inventory()
1130
>>> i1.add(InventoryFile('123', 'foo', ROOT_ID))
1131
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
1134
>>> i2.add(InventoryFile('123', 'foo', ROOT_ID))
1135
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
1139
if not isinstance(other, Inventory):
1140
return NotImplemented
1142
return self._byid == other._byid
1144
def __ne__(self, other):
1145
return not self.__eq__(other)
1148
raise ValueError('not hashable')
1150
def _iter_file_id_parents(self, file_id):
1151
"""Yield the parents of file_id up to the root."""
1152
while file_id is not None:
1154
ie = self._byid[file_id]
1156
raise BzrError("file_id {%s} not found in inventory" % file_id)
1158
file_id = ie.parent_id
1160
def get_idpath(self, file_id):
1161
"""Return a list of file_ids for the path to an entry.
1163
The list contains one element for each directory followed by
1164
the id of the file itself. So the length of the returned list
1165
is equal to the depth of the file in the tree, counting the
1166
root directory as depth 1.
1169
for parent in self._iter_file_id_parents(file_id):
1170
p.insert(0, parent.file_id)
1173
def id2path(self, file_id):
1174
"""Return as a string the path to file_id.
1177
>>> e = i.add(InventoryDirectory('src-id', 'src', ROOT_ID))
1178
>>> e = i.add(InventoryFile('foo-id', 'foo.c', parent_id='src-id'))
1179
>>> print i.id2path('foo-id')
1182
# get all names, skipping root
1183
return '/'.join(reversed(
1184
[parent.name for parent in
1185
self._iter_file_id_parents(file_id)][:-1]))
1187
def path2id(self, name):
1188
"""Walk down through directories to return entry of last component.
1190
names may be either a list of path components, or a single
1191
string, in which case it is automatically split.
1193
This returns the entry of the last component in the path,
1194
which may be either a file or a directory.
1196
Returns None IFF the path is not found.
1198
if isinstance(name, basestring):
1199
name = osutils.splitpath(name)
1201
# mutter("lookup path %r" % name)
1208
cie = parent.children[f]
1209
assert cie.name == f
1210
assert cie.parent_id == parent.file_id
1213
# or raise an error?
1216
return parent.file_id
1218
def has_filename(self, names):
1219
return bool(self.path2id(names))
1221
def has_id(self, file_id):
1222
return (file_id in self._byid)
1224
def remove_recursive_id(self, file_id):
1225
"""Remove file_id, and children, from the inventory.
1227
:param file_id: A file_id to remove.
1229
to_find_delete = [self._byid[file_id]]
1231
while to_find_delete:
1232
ie = to_find_delete.pop()
1233
to_delete.append(ie.file_id)
1234
if ie.kind == 'directory':
1235
to_find_delete.extend(ie.children.values())
1236
for file_id in reversed(to_delete):
1238
del self._byid[file_id]
1239
if ie.parent_id is not None:
1240
del self[ie.parent_id].children[ie.name]
1242
def rename(self, file_id, new_parent_id, new_name):
1243
"""Move a file within the inventory.
1245
This can change either the name, or the parent, or both.
1247
This does not move the working file."""
1248
if not is_valid_name(new_name):
1249
raise BzrError("not an acceptable filename: %r" % new_name)
1251
new_parent = self._byid[new_parent_id]
1252
if new_name in new_parent.children:
1253
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
1255
new_parent_idpath = self.get_idpath(new_parent_id)
1256
if file_id in new_parent_idpath:
1257
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
1258
% (self.id2path(file_id), self.id2path(new_parent_id)))
1260
file_ie = self._byid[file_id]
1261
old_parent = self._byid[file_ie.parent_id]
1263
# TODO: Don't leave things messed up if this fails
1265
del old_parent.children[file_ie.name]
1266
new_parent.children[new_name] = file_ie
1268
file_ie.name = new_name
1269
file_ie.parent_id = new_parent_id
1271
def is_root(self, file_id):
1272
return self.root is not None and file_id == self.root.file_id
1275
def make_entry(kind, name, parent_id, file_id=None):
1276
"""Create an inventory entry.
1278
:param kind: the type of inventory entry to create.
1279
:param name: the basename of the entry.
1280
:param parent_id: the parent_id of the entry.
1281
:param file_id: the file_id to use. if None, one will be created.
1284
file_id = workingtree.gen_file_id(name)
1286
norm_name, can_access = osutils.normalized_filename(name)
1287
if norm_name != name:
1291
# TODO: jam 20060701 This would probably be more useful
1292
# if the error was raised with the full path
1293
raise errors.InvalidNormalization(name)
1295
if kind == 'directory':
1296
return InventoryDirectory(file_id, name, parent_id)
1297
elif kind == 'file':
1298
return InventoryFile(file_id, name, parent_id)
1299
elif kind == 'symlink':
1300
return InventoryLink(file_id, name, parent_id)
1302
raise BzrError("unknown kind %r" % kind)
1307
def is_valid_name(name):
1309
if _NAME_RE is None:
1310
_NAME_RE = re.compile(r'^[^/\\]+$')
1312
return bool(_NAME_RE.match(name))
added
removed
bzrlib/inventory.py
