1
# Copyright (C) 2005, 2006 Canonical Ltd
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
# FIXME: This refactoring of the workingtree code doesn't seem to keep
18
# the WorkingTree's copy of the inventory in sync with the branch. The
19
# branch modifies its working inventory when it does a commit to make
20
# missing files permanently removed.
22
# TODO: Maybe also keep the full path of the entry, and the children?
23
# But those depend on its position within a particular inventory, and
24
# it would be nice not to need to hold the backpointer here.
26
# This should really be an id randomly assigned when the tree is
27
# created, but it's not for now.
39
from bzrlib import errors, osutils
40
from bzrlib.osutils import (pumpfile, quotefn, splitpath, joinpath,
41
pathjoin, sha_strings)
42
from bzrlib.errors import (NotVersionedError, InvalidEntryName,
43
BzrError, BzrCheckError, BinaryFile)
44
from bzrlib.trace import mutter
47
class InventoryEntry(object):
48
"""Description of a versioned file.
50
An InventoryEntry has the following fields, which are also
51
present in the XML inventory-entry element:
56
(within the parent directory)
59
file_id of the parent directory, or ROOT_ID
62
the revision_id in which this variation of this file was
66
Indicates that this file should be executable on systems
70
sha-1 of the text of the file
73
size in bytes of the text of the file
75
(reading a version 4 tree created a text_id field.)
80
>>> i.add(InventoryDirectory('123', 'src', ROOT_ID))
81
InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None)
82
>>> i.add(InventoryFile('2323', 'hello.c', parent_id='123'))
83
InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None)
84
>>> shouldbe = {0: '', 1: 'src', 2: pathjoin('src','hello.c')}
85
>>> for ix, j in enumerate(i.iter_entries()):
86
... print (j[0] == shouldbe[ix], j[1])
88
(True, RootEntry('TREE_ROOT', u'', parent_id=None, revision=None))
89
(True, InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None))
90
(True, InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None))
91
>>> i.add(InventoryFile('2323', 'bye.c', '123'))
92
Traceback (most recent call last):
94
BzrError: inventory already contains entry with id {2323}
95
>>> i.add(InventoryFile('2324', 'bye.c', '123'))
96
InventoryFile('2324', 'bye.c', parent_id='123', sha1=None, len=None)
97
>>> i.add(InventoryDirectory('2325', 'wibble', '123'))
98
InventoryDirectory('2325', 'wibble', parent_id='123', revision=None)
99
>>> i.path2id('src/wibble')
103
>>> i.add(InventoryFile('2326', 'wibble.c', '2325'))
104
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
106
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
107
>>> for path, entry in i.iter_entries():
109
... assert i.path2id(path)
117
>>> i.id2path('2326')
118
'src/wibble/wibble.c'
121
# Constants returned by describe_change()
123
# TODO: These should probably move to some kind of FileChangeDescription
124
# class; that's like what's inside a TreeDelta but we want to be able to
125
# generate them just for one file at a time.
127
MODIFIED_AND_RENAMED = 'modified and renamed'
131
def detect_changes(self, old_entry):
132
"""Return a (text_modified, meta_modified) from this to old_entry.
134
_read_tree_state must have been called on self and old_entry prior to
135
calling detect_changes.
139
def diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
140
output_to, reverse=False):
141
"""Perform a diff from this to to_entry.
143
text_diff will be used for textual difference calculation.
144
This is a template method, override _diff in child classes.
146
self._read_tree_state(tree.id2path(self.file_id), tree)
148
# cannot diff from one kind to another - you must do a removal
149
# and an addif they do not match.
150
assert self.kind == to_entry.kind
151
to_entry._read_tree_state(to_tree.id2path(to_entry.file_id),
153
self._diff(text_diff, from_label, tree, to_label, to_entry, to_tree,
156
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
157
output_to, reverse=False):
158
"""Perform a diff between two entries of the same kind."""
160
def find_previous_heads(self, previous_inventories,
161
versioned_file_store,
164
"""Return the revisions and entries that directly precede this.
166
Returned as a map from revision to inventory entry.
168
This is a map containing the file revisions in all parents
169
for which the file exists, and its revision is not a parent of
170
any other. If the file is new, the set will be empty.
172
:param versioned_file_store: A store where ancestry data on this
173
file id can be queried.
174
:param transaction: The transaction that queries to the versioned
175
file store should be completed under.
176
:param entry_vf: The entry versioned file, if its already available.
178
def get_ancestors(weave, entry):
179
return set(weave.get_ancestry(entry.revision))
180
# revision:ie mapping for each ie found in previous_inventories.
182
# revision:ie mapping with one revision for each head.
184
# revision: ancestor list for each head
186
# identify candidate head revision ids.
187
for inv in previous_inventories:
188
if self.file_id in inv:
189
ie = inv[self.file_id]
190
assert ie.file_id == self.file_id
191
if ie.revision in candidates:
192
# same revision value in two different inventories:
193
# correct possible inconsistencies:
194
# * there was a bug in revision updates with 'x' bit
197
if candidates[ie.revision].executable != ie.executable:
198
candidates[ie.revision].executable = False
199
ie.executable = False
200
except AttributeError:
202
# must now be the same.
203
assert candidates[ie.revision] == ie
205
# add this revision as a candidate.
206
candidates[ie.revision] = ie
208
# common case optimisation
209
if len(candidates) == 1:
210
# if there is only one candidate revision found
211
# then we can opening the versioned file to access ancestry:
212
# there cannot be any ancestors to eliminate when there is
213
# only one revision available.
214
heads[ie.revision] = ie
217
# eliminate ancestors amongst the available candidates:
218
# heads are those that are not an ancestor of any other candidate
219
# - this provides convergence at a per-file level.
220
for ie in candidates.values():
221
# may be an ancestor of a known head:
222
already_present = 0 != len(
223
[head for head in heads
224
if ie.revision in head_ancestors[head]])
226
# an ancestor of an analyzed candidate.
228
# not an ancestor of a known head:
229
# load the versioned file for this file id if needed
231
entry_vf = versioned_file_store.get_weave_or_empty(
232
self.file_id, transaction)
233
ancestors = get_ancestors(entry_vf, ie)
234
# may knock something else out:
235
check_heads = list(heads.keys())
236
for head in check_heads:
237
if head in ancestors:
238
# this previously discovered 'head' is not
239
# really a head - its an ancestor of the newly
242
head_ancestors[ie.revision] = ancestors
243
heads[ie.revision] = ie
246
def get_tar_item(self, root, dp, now, tree):
247
"""Get a tarfile item and a file stream for its content."""
248
item = tarfile.TarInfo(pathjoin(root, dp))
249
# TODO: would be cool to actually set it to the timestamp of the
250
# revision it was last changed
252
fileobj = self._put_in_tar(item, tree)
256
"""Return true if the object this entry represents has textual data.
258
Note that textual data includes binary content.
260
Also note that all entries get weave files created for them.
261
This attribute is primarily used when upgrading from old trees that
262
did not have the weave index for all inventory entries.
266
def __init__(self, file_id, name, parent_id, text_id=None):
267
"""Create an InventoryEntry
269
The filename must be a single component, relative to the
270
parent directory; it cannot be a whole path or relative name.
272
>>> e = InventoryFile('123', 'hello.c', ROOT_ID)
277
>>> e = InventoryFile('123', 'src/hello.c', ROOT_ID)
278
Traceback (most recent call last):
279
InvalidEntryName: Invalid entry name: src/hello.c
281
assert isinstance(name, basestring), name
282
if '/' in name or '\\' in name:
283
raise InvalidEntryName(name=name)
284
self.executable = False
286
self.text_sha1 = None
287
self.text_size = None
288
self.file_id = file_id
290
self.text_id = text_id
291
self.parent_id = parent_id
292
self.symlink_target = None
294
def kind_character(self):
295
"""Return a short kind indicator useful for appending to names."""
296
raise BzrError('unknown kind %r' % self.kind)
298
known_kinds = ('file', 'directory', 'symlink', 'root_directory')
300
def _put_in_tar(self, item, tree):
301
"""populate item for stashing in a tar, and return the content stream.
303
If no content is available, return None.
305
raise BzrError("don't know how to export {%s} of kind %r" %
306
(self.file_id, self.kind))
308
def put_on_disk(self, dest, dp, tree):
309
"""Create a representation of self on disk in the prefix dest.
311
This is a template method - implement _put_on_disk in subclasses.
313
fullpath = pathjoin(dest, dp)
314
self._put_on_disk(fullpath, tree)
315
# mutter(" export {%s} kind %s to %s", self.file_id,
316
# self.kind, fullpath)
318
def _put_on_disk(self, fullpath, tree):
319
"""Put this entry onto disk at fullpath, from tree tree."""
320
raise BzrError("don't know how to export {%s} of kind %r" % (self.file_id, self.kind))
322
def sorted_children(self):
323
return sorted(self.children.items())
326
def versionable_kind(kind):
327
return kind in ('file', 'directory', 'symlink')
329
def check(self, checker, rev_id, inv, tree):
330
"""Check this inventory entry is intact.
332
This is a template method, override _check for kind specific
335
:param checker: Check object providing context for the checks;
336
can be used to find out what parts of the repository have already
338
:param rev_id: Revision id from which this InventoryEntry was loaded.
339
Not necessarily the last-changed revision for this file.
340
:param inv: Inventory from which the entry was loaded.
341
:param tree: RevisionTree for this entry.
343
if self.parent_id is not None:
344
if not inv.has_id(self.parent_id):
345
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
346
% (self.parent_id, rev_id))
347
self._check(checker, rev_id, tree)
349
def _check(self, checker, rev_id, tree):
350
"""Check this inventory entry for kind specific errors."""
351
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
355
"""Clone this inventory entry."""
356
raise NotImplementedError
359
def describe_change(old_entry, new_entry):
360
"""Describe the change between old_entry and this.
362
This smells of being an InterInventoryEntry situation, but as its
363
the first one, we're making it a static method for now.
365
An entry with a different parent, or different name is considered
366
to be renamed. Reparenting is an internal detail.
367
Note that renaming the parent does not trigger a rename for the
370
# TODO: Perhaps return an object rather than just a string
371
if old_entry is new_entry:
372
# also the case of both being None
374
elif old_entry is None:
376
elif new_entry is None:
378
text_modified, meta_modified = new_entry.detect_changes(old_entry)
379
if text_modified or meta_modified:
383
# TODO 20060511 (mbp, rbc) factor out 'detect_rename' here.
384
if old_entry.parent_id != new_entry.parent_id:
386
elif old_entry.name != new_entry.name:
390
if renamed and not modified:
391
return InventoryEntry.RENAMED
392
if modified and not renamed:
394
if modified and renamed:
395
return InventoryEntry.MODIFIED_AND_RENAMED
399
return ("%s(%r, %r, parent_id=%r, revision=%r)"
400
% (self.__class__.__name__,
406
def snapshot(self, revision, path, previous_entries,
407
work_tree, commit_builder):
408
"""Make a snapshot of this entry which may or may not have changed.
410
This means that all its fields are populated, that it has its
411
text stored in the text store or weave.
413
# mutter('new parents of %s are %r', path, previous_entries)
414
self._read_tree_state(path, work_tree)
415
# TODO: Where should we determine whether to reuse a
416
# previous revision id or create a new revision? 20060606
417
if len(previous_entries) == 1:
418
# cannot be unchanged unless there is only one parent file rev.
419
parent_ie = previous_entries.values()[0]
420
if self._unchanged(parent_ie):
421
# mutter("found unchanged entry")
422
self.revision = parent_ie.revision
424
return self._snapshot_into_revision(revision, previous_entries,
425
work_tree, commit_builder)
427
def _snapshot_into_revision(self, revision, previous_entries, work_tree,
429
"""Record this revision unconditionally into a store.
431
The entry's last-changed revision property (`revision`) is updated to
432
that of the new revision.
434
:param revision: id of the new revision that is being recorded.
436
:returns: String description of the commit (e.g. "merged", "modified"), etc.
438
# mutter('new revision {%s} for {%s}', revision, self.file_id)
439
self.revision = revision
440
self._snapshot_text(previous_entries, work_tree, commit_builder)
442
def _snapshot_text(self, file_parents, work_tree, commit_builder):
443
"""Record the 'text' of this entry, whatever form that takes.
445
This default implementation simply adds an empty text.
447
raise NotImplementedError(self._snapshot_text)
449
def __eq__(self, other):
450
if not isinstance(other, InventoryEntry):
451
return NotImplemented
453
return ((self.file_id == other.file_id)
454
and (self.name == other.name)
455
and (other.symlink_target == self.symlink_target)
456
and (self.text_sha1 == other.text_sha1)
457
and (self.text_size == other.text_size)
458
and (self.text_id == other.text_id)
459
and (self.parent_id == other.parent_id)
460
and (self.kind == other.kind)
461
and (self.revision == other.revision)
462
and (self.executable == other.executable)
465
def __ne__(self, other):
466
return not (self == other)
469
raise ValueError('not hashable')
471
def _unchanged(self, previous_ie):
472
"""Has this entry changed relative to previous_ie.
474
This method should be overridden in child classes.
477
# different inv parent
478
if previous_ie.parent_id != self.parent_id:
481
elif previous_ie.name != self.name:
485
def _read_tree_state(self, path, work_tree):
486
"""Populate fields in the inventory entry from the given tree.
488
Note that this should be modified to be a noop on virtual trees
489
as all entries created there are prepopulated.
491
# TODO: Rather than running this manually, we should check the
492
# working sha1 and other expensive properties when they're
493
# first requested, or preload them if they're already known
494
pass # nothing to do by default
496
def _forget_tree_state(self):
500
class RootEntry(InventoryEntry):
502
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
503
'text_id', 'parent_id', 'children', 'executable',
504
'revision', 'symlink_target']
506
def _check(self, checker, rev_id, tree):
507
"""See InventoryEntry._check"""
509
def __init__(self, file_id):
510
self.file_id = file_id
512
self.kind = 'root_directory'
513
self.parent_id = None
517
def __eq__(self, other):
518
if not isinstance(other, RootEntry):
519
return NotImplemented
521
return (self.file_id == other.file_id) \
522
and (self.children == other.children)
525
class InventoryDirectory(InventoryEntry):
526
"""A directory in an inventory."""
528
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
529
'text_id', 'parent_id', 'children', 'executable',
530
'revision', 'symlink_target']
532
def _check(self, checker, rev_id, tree):
533
"""See InventoryEntry._check"""
534
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
535
raise BzrCheckError('directory {%s} has text in revision {%s}'
536
% (self.file_id, rev_id))
539
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
540
other.revision = self.revision
541
# note that children are *not* copied; they're pulled across when
545
def __init__(self, file_id, name, parent_id):
546
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
548
self.kind = 'directory'
550
def kind_character(self):
551
"""See InventoryEntry.kind_character."""
554
def _put_in_tar(self, item, tree):
555
"""See InventoryEntry._put_in_tar."""
556
item.type = tarfile.DIRTYPE
563
def _put_on_disk(self, fullpath, tree):
564
"""See InventoryEntry._put_on_disk."""
567
def _snapshot_text(self, file_parents, work_tree, commit_builder):
568
"""See InventoryEntry._snapshot_text."""
569
commit_builder.modified_directory(self.file_id, file_parents)
572
class InventoryFile(InventoryEntry):
573
"""A file in an inventory."""
575
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
576
'text_id', 'parent_id', 'children', 'executable',
577
'revision', 'symlink_target']
579
def _check(self, checker, tree_revision_id, tree):
580
"""See InventoryEntry._check"""
581
t = (self.file_id, self.revision)
582
if t in checker.checked_texts:
583
prev_sha = checker.checked_texts[t]
584
if prev_sha != self.text_sha1:
585
raise BzrCheckError('mismatched sha1 on {%s} in {%s}' %
586
(self.file_id, tree_revision_id))
588
checker.repeated_text_cnt += 1
591
if self.file_id not in checker.checked_weaves:
592
mutter('check weave {%s}', self.file_id)
593
w = tree.get_weave(self.file_id)
594
# Not passing a progress bar, because it creates a new
595
# progress, which overwrites the current progress,
596
# and doesn't look nice
598
checker.checked_weaves[self.file_id] = True
600
w = tree.get_weave(self.file_id)
602
mutter('check version {%s} of {%s}', tree_revision_id, self.file_id)
603
checker.checked_text_cnt += 1
604
# We can't check the length, because Weave doesn't store that
605
# information, and the whole point of looking at the weave's
606
# sha1sum is that we don't have to extract the text.
607
if self.text_sha1 != w.get_sha1(self.revision):
608
raise BzrCheckError('text {%s} version {%s} wrong sha1'
609
% (self.file_id, self.revision))
610
checker.checked_texts[t] = self.text_sha1
613
other = InventoryFile(self.file_id, self.name, self.parent_id)
614
other.executable = self.executable
615
other.text_id = self.text_id
616
other.text_sha1 = self.text_sha1
617
other.text_size = self.text_size
618
other.revision = self.revision
621
def detect_changes(self, old_entry):
622
"""See InventoryEntry.detect_changes."""
623
assert self.text_sha1 is not None
624
assert old_entry.text_sha1 is not None
625
text_modified = (self.text_sha1 != old_entry.text_sha1)
626
meta_modified = (self.executable != old_entry.executable)
627
return text_modified, meta_modified
629
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
630
output_to, reverse=False):
631
"""See InventoryEntry._diff."""
633
from_text = tree.get_file(self.file_id).readlines()
635
to_text = to_tree.get_file(to_entry.file_id).readlines()
639
text_diff(from_label, from_text,
640
to_label, to_text, output_to)
642
text_diff(to_label, to_text,
643
from_label, from_text, output_to)
646
label_pair = (to_label, from_label)
648
label_pair = (from_label, to_label)
649
print >> output_to, "Binary files %s and %s differ" % label_pair
652
"""See InventoryEntry.has_text."""
655
def __init__(self, file_id, name, parent_id):
656
super(InventoryFile, self).__init__(file_id, name, parent_id)
659
def kind_character(self):
660
"""See InventoryEntry.kind_character."""
663
def _put_in_tar(self, item, tree):
664
"""See InventoryEntry._put_in_tar."""
665
item.type = tarfile.REGTYPE
666
fileobj = tree.get_file(self.file_id)
667
item.size = self.text_size
668
if tree.is_executable(self.file_id):
674
def _put_on_disk(self, fullpath, tree):
675
"""See InventoryEntry._put_on_disk."""
676
pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
677
if tree.is_executable(self.file_id):
678
os.chmod(fullpath, 0755)
680
def _read_tree_state(self, path, work_tree):
681
"""See InventoryEntry._read_tree_state."""
682
self.text_sha1 = work_tree.get_file_sha1(self.file_id, path=path)
683
# FIXME: 20050930 probe for the text size when getting sha1
684
# in _read_tree_state
685
self.executable = work_tree.is_executable(self.file_id, path=path)
688
return ("%s(%r, %r, parent_id=%r, sha1=%r, len=%s)"
689
% (self.__class__.__name__,
696
def _forget_tree_state(self):
697
self.text_sha1 = None
699
def _snapshot_text(self, file_parents, work_tree, commit_builder):
700
"""See InventoryEntry._snapshot_text."""
701
def get_content_byte_lines():
702
return work_tree.get_file(self.file_id).readlines()
703
self.text_sha1, self.text_size = commit_builder.modified_file_text(
704
self.file_id, file_parents, get_content_byte_lines, self.text_sha1, self.text_size)
706
def _unchanged(self, previous_ie):
707
"""See InventoryEntry._unchanged."""
708
compatible = super(InventoryFile, self)._unchanged(previous_ie)
709
if self.text_sha1 != previous_ie.text_sha1:
712
# FIXME: 20050930 probe for the text size when getting sha1
713
# in _read_tree_state
714
self.text_size = previous_ie.text_size
715
if self.executable != previous_ie.executable:
720
class InventoryLink(InventoryEntry):
721
"""A file in an inventory."""
723
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
724
'text_id', 'parent_id', 'children', 'executable',
725
'revision', 'symlink_target']
727
def _check(self, checker, rev_id, tree):
728
"""See InventoryEntry._check"""
729
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
730
raise BzrCheckError('symlink {%s} has text in revision {%s}'
731
% (self.file_id, rev_id))
732
if self.symlink_target is None:
733
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
734
% (self.file_id, rev_id))
737
other = InventoryLink(self.file_id, self.name, self.parent_id)
738
other.symlink_target = self.symlink_target
739
other.revision = self.revision
742
def detect_changes(self, old_entry):
743
"""See InventoryEntry.detect_changes."""
744
# FIXME: which _modified field should we use ? RBC 20051003
745
text_modified = (self.symlink_target != old_entry.symlink_target)
747
mutter(" symlink target changed")
748
meta_modified = False
749
return text_modified, meta_modified
751
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
752
output_to, reverse=False):
753
"""See InventoryEntry._diff."""
754
from_text = self.symlink_target
755
if to_entry is not None:
756
to_text = to_entry.symlink_target
761
print >>output_to, '=== target changed %r => %r' % (from_text, to_text)
764
print >>output_to, '=== target was %r' % self.symlink_target
766
print >>output_to, '=== target is %r' % self.symlink_target
768
def __init__(self, file_id, name, parent_id):
769
super(InventoryLink, self).__init__(file_id, name, parent_id)
770
self.kind = 'symlink'
772
def kind_character(self):
773
"""See InventoryEntry.kind_character."""
776
def _put_in_tar(self, item, tree):
777
"""See InventoryEntry._put_in_tar."""
778
item.type = tarfile.SYMTYPE
782
item.linkname = self.symlink_target
785
def _put_on_disk(self, fullpath, tree):
786
"""See InventoryEntry._put_on_disk."""
788
os.symlink(self.symlink_target, fullpath)
790
raise BzrError("Failed to create symlink %r -> %r, error: %s" % (fullpath, self.symlink_target, e))
792
def _read_tree_state(self, path, work_tree):
793
"""See InventoryEntry._read_tree_state."""
794
self.symlink_target = work_tree.get_symlink_target(self.file_id)
796
def _forget_tree_state(self):
797
self.symlink_target = None
799
def _unchanged(self, previous_ie):
800
"""See InventoryEntry._unchanged."""
801
compatible = super(InventoryLink, self)._unchanged(previous_ie)
802
if self.symlink_target != previous_ie.symlink_target:
806
def _snapshot_text(self, file_parents, work_tree, commit_builder):
807
"""See InventoryEntry._snapshot_text."""
808
commit_builder.modified_link(
809
self.file_id, file_parents, self.symlink_target)
812
class Inventory(object):
813
"""Inventory of versioned files in a tree.
815
This describes which file_id is present at each point in the tree,
816
and possibly the SHA-1 or other information about the file.
817
Entries can be looked up either by path or by file_id.
819
The inventory represents a typical unix file tree, with
820
directories containing files and subdirectories. We never store
821
the full path to a file, because renaming a directory implicitly
822
moves all of its contents. This class internally maintains a
823
lookup tree that allows the children under a directory to be
826
InventoryEntry objects must not be modified after they are
827
inserted, other than through the Inventory API.
829
>>> inv = Inventory()
830
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
831
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
832
>>> inv['123-123'].name
835
May be treated as an iterator or set to look up file ids:
837
>>> bool(inv.path2id('hello.c'))
842
May also look up by name:
844
>>> [x[0] for x in inv.iter_entries()]
846
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
847
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
848
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT-12345678-12345678', sha1=None, len=None)
850
def __init__(self, root_id=ROOT_ID, revision_id=None):
851
"""Create or read an inventory.
853
If a working directory is specified, the inventory is read
854
from there. If the file is specified, read from that. If not,
855
the inventory is created empty.
857
The inventory is created with a default root directory, with
860
# We are letting Branch.create() create a unique inventory
861
# root id. Rather than generating a random one here.
863
# root_id = bzrlib.branch.gen_file_id('TREE_ROOT')
864
self.root = RootEntry(root_id)
865
# FIXME: this isn't ever used, changing it to self.revision may break
866
# things. TODO make everything use self.revision_id
867
self.revision_id = revision_id
868
self._byid = {self.root.file_id: self.root}
871
# TODO: jam 20051218 Should copy also copy the revision_id?
872
entries = self.iter_entries()
873
other = Inventory(entries.next()[1].file_id)
874
# copy recursively so we know directories will be added before
875
# their children. There are more efficient ways than this...
876
for path, entry in entries():
877
other.add(entry.copy())
881
return iter(self._byid)
884
"""Returns number of entries."""
885
return len(self._byid)
887
def iter_entries(self, from_dir=None):
888
"""Return (path, entry) pairs, in order by name."""
893
elif isinstance(from_dir, basestring):
894
from_dir = self._byid[from_dir]
896
# unrolling the recursive called changed the time from
897
# 440ms/663ms (inline/total) to 116ms/116ms
898
children = from_dir.children.items()
900
children = collections.deque(children)
901
stack = [(u'', children)]
903
from_dir_relpath, children = stack[-1]
906
name, ie = children.popleft()
908
# we know that from_dir_relpath never ends in a slash
909
# and 'f' doesn't begin with one, we can do a string op, rather
910
# than the checks of pathjoin(), though this means that all paths
912
path = from_dir_relpath + '/' + name
916
if ie.kind != 'directory':
919
# But do this child first
920
new_children = ie.children.items()
922
new_children = collections.deque(new_children)
923
stack.append((path, new_children))
924
# Break out of inner loop, so that we start outer loop with child
927
# if we finished all children, pop it off the stack
930
def iter_entries_by_dir(self, from_dir=None):
931
"""Iterate over the entries in a directory first order.
933
This returns all entries for a directory before returning
934
the entries for children of a directory. This is not
935
lexicographically sorted order, and is a hybrid between
936
depth-first and breadth-first.
938
:return: This yields (path, entry) pairs
940
# TODO? Perhaps this should return the from_dir so that the root is
941
# yielded? or maybe an option?
946
elif isinstance(from_dir, basestring):
947
from_dir = self._byid[from_dir]
949
stack = [(u'', from_dir)]
951
cur_relpath, cur_dir = stack.pop()
954
for child_name, child_ie in sorted(cur_dir.children.iteritems()):
956
child_relpath = cur_relpath + child_name
958
yield child_relpath, child_ie
960
if child_ie.kind == 'directory':
961
child_dirs.append((child_relpath+'/', child_ie))
962
stack.extend(reversed(child_dirs))
965
"""Return list of (path, ie) for all entries except the root.
967
This may be faster than iter_entries.
970
def descend(dir_ie, dir_path):
971
kids = dir_ie.children.items()
973
for name, ie in kids:
974
child_path = pathjoin(dir_path, name)
975
accum.append((child_path, ie))
976
if ie.kind == 'directory':
977
descend(ie, child_path)
979
descend(self.root, u'')
982
def directories(self):
983
"""Return (path, entry) pairs for all directories, including the root.
986
def descend(parent_ie, parent_path):
987
accum.append((parent_path, parent_ie))
989
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
992
for name, child_ie in kids:
993
child_path = pathjoin(parent_path, name)
994
descend(child_ie, child_path)
995
descend(self.root, u'')
998
def __contains__(self, file_id):
999
"""True if this entry contains a file with given id.
1001
>>> inv = Inventory()
1002
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1003
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
1009
return file_id in self._byid
1011
def __getitem__(self, file_id):
1012
"""Return the entry for given file_id.
1014
>>> inv = Inventory()
1015
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
1016
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
1017
>>> inv['123123'].name
1021
return self._byid[file_id]
1024
raise BzrError("can't look up file_id None")
1026
raise BzrError("file_id {%s} not in inventory" % file_id)
1028
def get_file_kind(self, file_id):
1029
return self._byid[file_id].kind
1031
def get_child(self, parent_id, filename):
1032
return self[parent_id].children.get(filename)
1034
def add(self, entry):
1035
"""Add entry to inventory.
1037
To add a file to a branch ready to be committed, use Branch.add,
1040
Returns the new entry object.
1042
if entry.file_id in self._byid:
1043
raise BzrError("inventory already contains entry with id {%s}" % entry.file_id)
1045
if entry.parent_id == ROOT_ID or entry.parent_id is None:
1046
entry.parent_id = self.root.file_id
1049
parent = self._byid[entry.parent_id]
1051
raise BzrError("parent_id {%s} not in inventory" % entry.parent_id)
1053
if entry.name in parent.children:
1054
raise BzrError("%s is already versioned" %
1055
pathjoin(self.id2path(parent.file_id), entry.name))
1057
self._byid[entry.file_id] = entry
1058
parent.children[entry.name] = entry
1061
def add_path(self, relpath, kind, file_id=None, parent_id=None):
1062
"""Add entry from a path.
1064
The immediate parent must already be versioned.
1066
Returns the new entry object."""
1068
parts = osutils.splitpath(relpath)
1072
file_id = bzrlib.workingtree.gen_root_id()
1073
self.root = RootEntry(file_id)
1074
self._byid = {self.root.file_id: self.root}
1077
parent_path = parts[:-1]
1078
parent_id = self.path2id(parent_path)
1079
if parent_id is None:
1080
raise NotVersionedError(path=parent_path)
1081
ie = make_entry(kind, parts[-1], parent_id, file_id)
1084
def __delitem__(self, file_id):
1085
"""Remove entry by id.
1087
>>> inv = Inventory()
1088
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1089
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
1098
assert ie.parent_id is None or \
1099
self[ie.parent_id].children[ie.name] == ie
1101
del self._byid[file_id]
1102
if ie.parent_id is not None:
1103
del self[ie.parent_id].children[ie.name]
1105
def __eq__(self, other):
1106
"""Compare two sets by comparing their contents.
1108
>>> i1 = Inventory()
1109
>>> i2 = Inventory()
1112
>>> i1.add(InventoryFile('123', 'foo', ROOT_ID))
1113
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
1116
>>> i2.add(InventoryFile('123', 'foo', ROOT_ID))
1117
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
1121
if not isinstance(other, Inventory):
1122
return NotImplemented
1124
return self._byid == other._byid
1126
def __ne__(self, other):
1127
return not self.__eq__(other)
1130
raise ValueError('not hashable')
1132
def _iter_file_id_parents(self, file_id):
1133
"""Yield the parents of file_id up to the root."""
1134
while file_id is not None:
1136
ie = self._byid[file_id]
1138
raise BzrError("file_id {%s} not found in inventory" % file_id)
1140
file_id = ie.parent_id
1142
def get_idpath(self, file_id):
1143
"""Return a list of file_ids for the path to an entry.
1145
The list contains one element for each directory followed by
1146
the id of the file itself. So the length of the returned list
1147
is equal to the depth of the file in the tree, counting the
1148
root directory as depth 1.
1151
for parent in self._iter_file_id_parents(file_id):
1152
p.insert(0, parent.file_id)
1155
def id2path(self, file_id):
1156
"""Return as a string the path to file_id.
1159
>>> e = i.add(InventoryDirectory('src-id', 'src', ROOT_ID))
1160
>>> e = i.add(InventoryFile('foo-id', 'foo.c', parent_id='src-id'))
1161
>>> print i.id2path('foo-id')
1164
# get all names, skipping root
1165
return '/'.join(reversed(
1166
[parent.name for parent in
1167
self._iter_file_id_parents(file_id)][:-1]))
1169
def path2id(self, name):
1170
"""Walk down through directories to return entry of last component.
1172
names may be either a list of path components, or a single
1173
string, in which case it is automatically split.
1175
This returns the entry of the last component in the path,
1176
which may be either a file or a directory.
1178
Returns None IFF the path is not found.
1180
if isinstance(name, types.StringTypes):
1181
name = splitpath(name)
1183
# mutter("lookup path %r" % name)
1188
cie = parent.children[f]
1189
assert cie.name == f
1190
assert cie.parent_id == parent.file_id
1193
# or raise an error?
1196
return parent.file_id
1198
def has_filename(self, names):
1199
return bool(self.path2id(names))
1201
def has_id(self, file_id):
1202
return self._byid.has_key(file_id)
1204
def rename(self, file_id, new_parent_id, new_name):
1205
"""Move a file within the inventory.
1207
This can change either the name, or the parent, or both.
1209
This does not move the working file."""
1210
if not is_valid_name(new_name):
1211
raise BzrError("not an acceptable filename: %r" % new_name)
1213
new_parent = self._byid[new_parent_id]
1214
if new_name in new_parent.children:
1215
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
1217
new_parent_idpath = self.get_idpath(new_parent_id)
1218
if file_id in new_parent_idpath:
1219
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
1220
% (self.id2path(file_id), self.id2path(new_parent_id)))
1222
file_ie = self._byid[file_id]
1223
old_parent = self._byid[file_ie.parent_id]
1225
# TODO: Don't leave things messed up if this fails
1227
del old_parent.children[file_ie.name]
1228
new_parent.children[new_name] = file_ie
1230
file_ie.name = new_name
1231
file_ie.parent_id = new_parent_id
1234
def make_entry(kind, name, parent_id, file_id=None):
1235
"""Create an inventory entry.
1237
:param kind: the type of inventory entry to create.
1238
:param name: the basename of the entry.
1239
:param parent_id: the parent_id of the entry.
1240
:param file_id: the file_id to use. if None, one will be created.
1243
file_id = bzrlib.workingtree.gen_file_id(name)
1245
norm_name, can_access = osutils.normalized_filename(name)
1246
if norm_name != name:
1250
# TODO: jam 20060701 This would probably be more useful
1251
# if the error was raised with the full path
1252
raise errors.InvalidNormalization(name)
1254
if kind == 'directory':
1255
return InventoryDirectory(file_id, name, parent_id)
1256
elif kind == 'file':
1257
return InventoryFile(file_id, name, parent_id)
1258
elif kind == 'symlink':
1259
return InventoryLink(file_id, name, parent_id)
1261
raise BzrError("unknown kind %r" % kind)
1266
def is_valid_name(name):
1268
if _NAME_RE is None:
1269
_NAME_RE = re.compile(r'^[^/\\]+$')
1271
return bool(_NAME_RE.match(name))
added
removed
bzrlib/inventory.py
