1
# Copyright (C) 2005, 2006 Canonical Ltd
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
# FIXME: This refactoring of the workingtree code doesn't seem to keep
18
# the WorkingTree's copy of the inventory in sync with the branch. The
19
# branch modifies its working inventory when it does a commit to make
20
# missing files permanently removed.
22
# TODO: Maybe also keep the full path of the entry, and the children?
23
# But those depend on its position within a particular inventory, and
24
# it would be nice not to need to hold the backpointer here.
26
# This should really be an id randomly assigned when the tree is
27
# created, but it's not for now.
39
from bzrlib.osutils import (pumpfile, quotefn, splitpath, joinpath,
40
pathjoin, sha_strings)
41
from bzrlib.errors import (NotVersionedError, InvalidEntryName,
42
BzrError, BzrCheckError, BinaryFile)
43
from bzrlib.trace import mutter
46
class InventoryEntry(object):
47
"""Description of a versioned file.
49
An InventoryEntry has the following fields, which are also
50
present in the XML inventory-entry element:
55
(within the parent directory)
58
file_id of the parent directory, or ROOT_ID
61
the revision_id in which this variation of this file was
65
Indicates that this file should be executable on systems
69
sha-1 of the text of the file
72
size in bytes of the text of the file
74
(reading a version 4 tree created a text_id field.)
79
>>> i.add(InventoryDirectory('123', 'src', ROOT_ID))
80
InventoryDirectory('123', 'src', parent_id='TREE_ROOT')
81
>>> i.add(InventoryFile('2323', 'hello.c', parent_id='123'))
82
InventoryFile('2323', 'hello.c', parent_id='123')
83
>>> shouldbe = {0: 'src', 1: pathjoin('src','hello.c')}
84
>>> for ix, j in enumerate(i.iter_entries()):
85
... print (j[0] == shouldbe[ix], j[1])
87
(True, InventoryDirectory('123', 'src', parent_id='TREE_ROOT'))
88
(True, InventoryFile('2323', 'hello.c', parent_id='123'))
89
>>> i.add(InventoryFile('2323', 'bye.c', '123'))
90
Traceback (most recent call last):
92
BzrError: inventory already contains entry with id {2323}
93
>>> i.add(InventoryFile('2324', 'bye.c', '123'))
94
InventoryFile('2324', 'bye.c', parent_id='123')
95
>>> i.add(InventoryDirectory('2325', 'wibble', '123'))
96
InventoryDirectory('2325', 'wibble', parent_id='123')
97
>>> i.path2id('src/wibble')
101
>>> i.add(InventoryFile('2326', 'wibble.c', '2325'))
102
InventoryFile('2326', 'wibble.c', parent_id='2325')
104
InventoryFile('2326', 'wibble.c', parent_id='2325')
105
>>> for path, entry in i.iter_entries():
107
... assert i.path2id(path)
114
>>> i.id2path('2326')
115
'src/wibble/wibble.c'
118
# Constants returned by describe_change()
120
# TODO: These should probably move to some kind of FileChangeDescription
121
# class; that's like what's inside a TreeDelta but we want to be able to
122
# generate them just for one file at a time.
124
MODIFIED_AND_RENAMED = 'modified and renamed'
128
def detect_changes(self, old_entry):
129
"""Return a (text_modified, meta_modified) from this to old_entry.
131
_read_tree_state must have been called on self and old_entry prior to
132
calling detect_changes.
136
def diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
137
output_to, reverse=False):
138
"""Perform a diff from this to to_entry.
140
text_diff will be used for textual difference calculation.
141
This is a template method, override _diff in child classes.
143
self._read_tree_state(tree.id2path(self.file_id), tree)
145
# cannot diff from one kind to another - you must do a removal
146
# and an addif they do not match.
147
assert self.kind == to_entry.kind
148
to_entry._read_tree_state(to_tree.id2path(to_entry.file_id),
150
self._diff(text_diff, from_label, tree, to_label, to_entry, to_tree,
153
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
154
output_to, reverse=False):
155
"""Perform a diff between two entries of the same kind."""
157
def find_previous_heads(self, previous_inventories,
158
versioned_file_store,
161
"""Return the revisions and entries that directly precede this.
163
Returned as a map from revision to inventory entry.
165
This is a map containing the file revisions in all parents
166
for which the file exists, and its revision is not a parent of
167
any other. If the file is new, the set will be empty.
169
:param versioned_file_store: A store where ancestry data on this
170
file id can be queried.
171
:param transaction: The transaction that queries to the versioned
172
file store should be completed under.
173
:param entry_vf: The entry versioned file, if its already available.
175
def get_ancestors(weave, entry):
176
return set(weave.get_ancestry(entry.revision))
177
# revision:ie mapping for each ie found in previous_inventories.
179
# revision:ie mapping with one revision for each head.
181
# revision: ancestor list for each head
183
# identify candidate head revision ids.
184
for inv in previous_inventories:
185
if self.file_id in inv:
186
ie = inv[self.file_id]
187
assert ie.file_id == self.file_id
188
if ie.revision in candidates:
189
# same revision value in two different inventories:
190
# correct possible inconsistencies:
191
# * there was a bug in revision updates with 'x' bit
194
if candidates[ie.revision].executable != ie.executable:
195
candidates[ie.revision].executable = False
196
ie.executable = False
197
except AttributeError:
199
# must now be the same.
200
assert candidates[ie.revision] == ie
202
# add this revision as a candidate.
203
candidates[ie.revision] = ie
205
# common case optimisation
206
if len(candidates) == 1:
207
# if there is only one candidate revision found
208
# then we can opening the versioned file to access ancestry:
209
# there cannot be any ancestors to eliminate when there is
210
# only one revision available.
211
heads[ie.revision] = ie
214
# eliminate ancestors amongst the available candidates:
215
# heads are those that are not an ancestor of any other candidate
216
# - this provides convergence at a per-file level.
217
for ie in candidates.values():
218
# may be an ancestor of a known head:
219
already_present = 0 != len(
220
[head for head in heads
221
if ie.revision in head_ancestors[head]])
223
# an ancestor of an analyzed candidate.
225
# not an ancestor of a known head:
226
# load the versioned file for this file id if needed
228
entry_vf = versioned_file_store.get_weave_or_empty(
229
self.file_id, transaction)
230
ancestors = get_ancestors(entry_vf, ie)
231
# may knock something else out:
232
check_heads = list(heads.keys())
233
for head in check_heads:
234
if head in ancestors:
235
# this previously discovered 'head' is not
236
# really a head - its an ancestor of the newly
239
head_ancestors[ie.revision] = ancestors
240
heads[ie.revision] = ie
243
def get_tar_item(self, root, dp, now, tree):
244
"""Get a tarfile item and a file stream for its content."""
245
item = tarfile.TarInfo(pathjoin(root, dp))
246
# TODO: would be cool to actually set it to the timestamp of the
247
# revision it was last changed
249
fileobj = self._put_in_tar(item, tree)
253
"""Return true if the object this entry represents has textual data.
255
Note that textual data includes binary content.
257
Also note that all entries get weave files created for them.
258
This attribute is primarily used when upgrading from old trees that
259
did not have the weave index for all inventory entries.
263
def __init__(self, file_id, name, parent_id, text_id=None):
264
"""Create an InventoryEntry
266
The filename must be a single component, relative to the
267
parent directory; it cannot be a whole path or relative name.
269
>>> e = InventoryFile('123', 'hello.c', ROOT_ID)
274
>>> e = InventoryFile('123', 'src/hello.c', ROOT_ID)
275
Traceback (most recent call last):
276
InvalidEntryName: Invalid entry name: src/hello.c
278
assert isinstance(name, basestring), name
279
if '/' in name or '\\' in name:
280
raise InvalidEntryName(name=name)
281
self.executable = False
283
self.text_sha1 = None
284
self.text_size = None
285
self.file_id = file_id
287
self.text_id = text_id
288
self.parent_id = parent_id
289
self.symlink_target = None
291
def kind_character(self):
292
"""Return a short kind indicator useful for appending to names."""
293
raise BzrError('unknown kind %r' % self.kind)
295
known_kinds = ('file', 'directory', 'symlink', 'root_directory')
297
def _put_in_tar(self, item, tree):
298
"""populate item for stashing in a tar, and return the content stream.
300
If no content is available, return None.
302
raise BzrError("don't know how to export {%s} of kind %r" %
303
(self.file_id, self.kind))
305
def put_on_disk(self, dest, dp, tree):
306
"""Create a representation of self on disk in the prefix dest.
308
This is a template method - implement _put_on_disk in subclasses.
310
fullpath = pathjoin(dest, dp)
311
self._put_on_disk(fullpath, tree)
312
mutter(" export {%s} kind %s to %s", self.file_id,
315
def _put_on_disk(self, fullpath, tree):
316
"""Put this entry onto disk at fullpath, from tree tree."""
317
raise BzrError("don't know how to export {%s} of kind %r" % (self.file_id, self.kind))
319
def sorted_children(self):
320
return sorted(self.children.items())
323
def versionable_kind(kind):
324
return kind in ('file', 'directory', 'symlink')
326
def check(self, checker, rev_id, inv, tree):
327
"""Check this inventory entry is intact.
329
This is a template method, override _check for kind specific
332
:param checker: Check object providing context for the checks;
333
can be used to find out what parts of the repository have already
335
:param rev_id: Revision id from which this InventoryEntry was loaded.
336
Not necessarily the last-changed revision for this file.
337
:param inv: Inventory from which the entry was loaded.
338
:param tree: RevisionTree for this entry.
340
if self.parent_id is not None:
341
if not inv.has_id(self.parent_id):
342
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
343
% (self.parent_id, rev_id))
344
self._check(checker, rev_id, tree)
346
def _check(self, checker, rev_id, tree):
347
"""Check this inventory entry for kind specific errors."""
348
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
352
"""Clone this inventory entry."""
353
raise NotImplementedError
356
def describe_change(old_entry, new_entry):
357
"""Describe the change between old_entry and this.
359
This smells of being an InterInventoryEntry situation, but as its
360
the first one, we're making it a static method for now.
362
An entry with a different parent, or different name is considered
363
to be renamed. Reparenting is an internal detail.
364
Note that renaming the parent does not trigger a rename for the
367
# TODO: Perhaps return an object rather than just a string
368
if old_entry is new_entry:
369
# also the case of both being None
371
elif old_entry is None:
373
elif new_entry is None:
375
text_modified, meta_modified = new_entry.detect_changes(old_entry)
376
if text_modified or meta_modified:
380
# TODO 20060511 (mbp, rbc) factor out 'detect_rename' here.
381
if old_entry.parent_id != new_entry.parent_id:
383
elif old_entry.name != new_entry.name:
387
if renamed and not modified:
388
return InventoryEntry.RENAMED
389
if modified and not renamed:
391
if modified and renamed:
392
return InventoryEntry.MODIFIED_AND_RENAMED
396
return ("%s(%r, %r, parent_id=%r)"
397
% (self.__class__.__name__,
402
def snapshot(self, revision, path, previous_entries,
403
work_tree, commit_builder):
404
"""Make a snapshot of this entry which may or may not have changed.
406
This means that all its fields are populated, that it has its
407
text stored in the text store or weave.
409
mutter('new parents of %s are %r', path, previous_entries)
410
self._read_tree_state(path, work_tree)
411
# TODO: Where should we determine whether to reuse a
412
# previous revision id or create a new revision? 20060606
413
if len(previous_entries) == 1:
414
# cannot be unchanged unless there is only one parent file rev.
415
parent_ie = previous_entries.values()[0]
416
if self._unchanged(parent_ie):
417
mutter("found unchanged entry")
418
self.revision = parent_ie.revision
420
return self._snapshot_into_revision(revision, previous_entries,
421
work_tree, commit_builder)
423
def _snapshot_into_revision(self, revision, previous_entries, work_tree,
425
"""Record this revision unconditionally into a store.
427
The entry's last-changed revision property (`revision`) is updated to
428
that of the new revision.
430
:param revision: id of the new revision that is being recorded.
432
:returns: String description of the commit (e.g. "merged", "modified"), etc.
434
mutter('new revision {%s} for {%s}', revision, self.file_id)
435
self.revision = revision
436
self._snapshot_text(previous_entries, work_tree, commit_builder)
438
def _snapshot_text(self, file_parents, work_tree, commit_builder):
439
"""Record the 'text' of this entry, whatever form that takes.
441
This default implementation simply adds an empty text.
443
raise NotImplementedError(self._snapshot_text)
445
def __eq__(self, other):
446
if not isinstance(other, InventoryEntry):
447
return NotImplemented
449
return ((self.file_id == other.file_id)
450
and (self.name == other.name)
451
and (other.symlink_target == self.symlink_target)
452
and (self.text_sha1 == other.text_sha1)
453
and (self.text_size == other.text_size)
454
and (self.text_id == other.text_id)
455
and (self.parent_id == other.parent_id)
456
and (self.kind == other.kind)
457
and (self.revision == other.revision)
458
and (self.executable == other.executable)
461
def __ne__(self, other):
462
return not (self == other)
465
raise ValueError('not hashable')
467
def _unchanged(self, previous_ie):
468
"""Has this entry changed relative to previous_ie.
470
This method should be overridden in child classes.
473
# different inv parent
474
if previous_ie.parent_id != self.parent_id:
477
elif previous_ie.name != self.name:
481
def _read_tree_state(self, path, work_tree):
482
"""Populate fields in the inventory entry from the given tree.
484
Note that this should be modified to be a noop on virtual trees
485
as all entries created there are prepopulated.
487
# TODO: Rather than running this manually, we should check the
488
# working sha1 and other expensive properties when they're
489
# first requested, or preload them if they're already known
490
pass # nothing to do by default
492
def _forget_tree_state(self):
496
class RootEntry(InventoryEntry):
498
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
499
'text_id', 'parent_id', 'children', 'executable',
500
'revision', 'symlink_target']
502
def _check(self, checker, rev_id, tree):
503
"""See InventoryEntry._check"""
505
def __init__(self, file_id):
506
self.file_id = file_id
508
self.kind = 'root_directory'
509
self.parent_id = None
512
def __eq__(self, other):
513
if not isinstance(other, RootEntry):
514
return NotImplemented
516
return (self.file_id == other.file_id) \
517
and (self.children == other.children)
520
class InventoryDirectory(InventoryEntry):
521
"""A directory in an inventory."""
523
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
524
'text_id', 'parent_id', 'children', 'executable',
525
'revision', 'symlink_target']
527
def _check(self, checker, rev_id, tree):
528
"""See InventoryEntry._check"""
529
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
530
raise BzrCheckError('directory {%s} has text in revision {%s}'
531
% (self.file_id, rev_id))
534
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
535
other.revision = self.revision
536
# note that children are *not* copied; they're pulled across when
540
def __init__(self, file_id, name, parent_id):
541
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
543
self.kind = 'directory'
545
def kind_character(self):
546
"""See InventoryEntry.kind_character."""
549
def _put_in_tar(self, item, tree):
550
"""See InventoryEntry._put_in_tar."""
551
item.type = tarfile.DIRTYPE
558
def _put_on_disk(self, fullpath, tree):
559
"""See InventoryEntry._put_on_disk."""
562
def _snapshot_text(self, file_parents, work_tree, commit_builder):
563
"""See InventoryEntry._snapshot_text."""
564
commit_builder.modified_directory(self.file_id, file_parents)
567
class InventoryFile(InventoryEntry):
568
"""A file in an inventory."""
570
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
571
'text_id', 'parent_id', 'children', 'executable',
572
'revision', 'symlink_target']
574
def _check(self, checker, tree_revision_id, tree):
575
"""See InventoryEntry._check"""
576
t = (self.file_id, self.revision)
577
if t in checker.checked_texts:
578
prev_sha = checker.checked_texts[t]
579
if prev_sha != self.text_sha1:
580
raise BzrCheckError('mismatched sha1 on {%s} in {%s}' %
581
(self.file_id, tree_revision_id))
583
checker.repeated_text_cnt += 1
586
if self.file_id not in checker.checked_weaves:
587
mutter('check weave {%s}', self.file_id)
588
w = tree.get_weave(self.file_id)
589
# Not passing a progress bar, because it creates a new
590
# progress, which overwrites the current progress,
591
# and doesn't look nice
593
checker.checked_weaves[self.file_id] = True
595
w = tree.get_weave(self.file_id)
597
mutter('check version {%s} of {%s}', tree_revision_id, self.file_id)
598
checker.checked_text_cnt += 1
599
# We can't check the length, because Weave doesn't store that
600
# information, and the whole point of looking at the weave's
601
# sha1sum is that we don't have to extract the text.
602
if self.text_sha1 != w.get_sha1(self.revision):
603
raise BzrCheckError('text {%s} version {%s} wrong sha1'
604
% (self.file_id, self.revision))
605
checker.checked_texts[t] = self.text_sha1
608
other = InventoryFile(self.file_id, self.name, self.parent_id)
609
other.executable = self.executable
610
other.text_id = self.text_id
611
other.text_sha1 = self.text_sha1
612
other.text_size = self.text_size
613
other.revision = self.revision
616
def detect_changes(self, old_entry):
617
"""See InventoryEntry.detect_changes."""
618
assert self.text_sha1 is not None
619
assert old_entry.text_sha1 is not None
620
text_modified = (self.text_sha1 != old_entry.text_sha1)
621
meta_modified = (self.executable != old_entry.executable)
622
return text_modified, meta_modified
624
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
625
output_to, reverse=False):
626
"""See InventoryEntry._diff."""
628
from_text = tree.get_file(self.file_id).readlines()
630
to_text = to_tree.get_file(to_entry.file_id).readlines()
634
text_diff(from_label, from_text,
635
to_label, to_text, output_to)
637
text_diff(to_label, to_text,
638
from_label, from_text, output_to)
641
label_pair = (to_label, from_label)
643
label_pair = (from_label, to_label)
644
print >> output_to, "Binary files %s and %s differ" % label_pair
647
"""See InventoryEntry.has_text."""
650
def __init__(self, file_id, name, parent_id):
651
super(InventoryFile, self).__init__(file_id, name, parent_id)
654
def kind_character(self):
655
"""See InventoryEntry.kind_character."""
658
def _put_in_tar(self, item, tree):
659
"""See InventoryEntry._put_in_tar."""
660
item.type = tarfile.REGTYPE
661
fileobj = tree.get_file(self.file_id)
662
item.size = self.text_size
663
if tree.is_executable(self.file_id):
669
def _put_on_disk(self, fullpath, tree):
670
"""See InventoryEntry._put_on_disk."""
671
pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
672
if tree.is_executable(self.file_id):
673
os.chmod(fullpath, 0755)
675
def _read_tree_state(self, path, work_tree):
676
"""See InventoryEntry._read_tree_state."""
677
self.text_sha1 = work_tree.get_file_sha1(self.file_id, path=path)
678
# FIXME: 20050930 probe for the text size when getting sha1
679
# in _read_tree_state
680
self.executable = work_tree.is_executable(self.file_id, path=path)
682
def _forget_tree_state(self):
683
self.text_sha1 = None
684
self.executable = None
686
def _snapshot_text(self, file_parents, work_tree, commit_builder):
687
"""See InventoryEntry._snapshot_text."""
688
def get_content_byte_lines():
689
return work_tree.get_file(self.file_id).readlines()
690
self.text_sha1, self.text_size = commit_builder.modified_file_text(
691
self.file_id, file_parents, get_content_byte_lines, self.text_sha1, self.text_size)
693
def _unchanged(self, previous_ie):
694
"""See InventoryEntry._unchanged."""
695
compatible = super(InventoryFile, self)._unchanged(previous_ie)
696
if self.text_sha1 != previous_ie.text_sha1:
699
# FIXME: 20050930 probe for the text size when getting sha1
700
# in _read_tree_state
701
self.text_size = previous_ie.text_size
702
if self.executable != previous_ie.executable:
707
class InventoryLink(InventoryEntry):
708
"""A file in an inventory."""
710
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
711
'text_id', 'parent_id', 'children', 'executable',
712
'revision', 'symlink_target']
714
def _check(self, checker, rev_id, tree):
715
"""See InventoryEntry._check"""
716
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
717
raise BzrCheckError('symlink {%s} has text in revision {%s}'
718
% (self.file_id, rev_id))
719
if self.symlink_target is None:
720
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
721
% (self.file_id, rev_id))
724
other = InventoryLink(self.file_id, self.name, self.parent_id)
725
other.symlink_target = self.symlink_target
726
other.revision = self.revision
729
def detect_changes(self, old_entry):
730
"""See InventoryEntry.detect_changes."""
731
# FIXME: which _modified field should we use ? RBC 20051003
732
text_modified = (self.symlink_target != old_entry.symlink_target)
734
mutter(" symlink target changed")
735
meta_modified = False
736
return text_modified, meta_modified
738
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
739
output_to, reverse=False):
740
"""See InventoryEntry._diff."""
741
from_text = self.symlink_target
742
if to_entry is not None:
743
to_text = to_entry.symlink_target
748
print >>output_to, '=== target changed %r => %r' % (from_text, to_text)
751
print >>output_to, '=== target was %r' % self.symlink_target
753
print >>output_to, '=== target is %r' % self.symlink_target
755
def __init__(self, file_id, name, parent_id):
756
super(InventoryLink, self).__init__(file_id, name, parent_id)
757
self.kind = 'symlink'
759
def kind_character(self):
760
"""See InventoryEntry.kind_character."""
763
def _put_in_tar(self, item, tree):
764
"""See InventoryEntry._put_in_tar."""
765
item.type = tarfile.SYMTYPE
769
item.linkname = self.symlink_target
772
def _put_on_disk(self, fullpath, tree):
773
"""See InventoryEntry._put_on_disk."""
775
os.symlink(self.symlink_target, fullpath)
777
raise BzrError("Failed to create symlink %r -> %r, error: %s" % (fullpath, self.symlink_target, e))
779
def _read_tree_state(self, path, work_tree):
780
"""See InventoryEntry._read_tree_state."""
781
self.symlink_target = work_tree.get_symlink_target(self.file_id)
783
def _forget_tree_state(self):
784
self.symlink_target = None
786
def _unchanged(self, previous_ie):
787
"""See InventoryEntry._unchanged."""
788
compatible = super(InventoryLink, self)._unchanged(previous_ie)
789
if self.symlink_target != previous_ie.symlink_target:
793
def _snapshot_text(self, file_parents, work_tree, commit_builder):
794
"""See InventoryEntry._snapshot_text."""
795
commit_builder.modified_link(
796
self.file_id, file_parents, self.symlink_target)
799
class Inventory(object):
800
"""Inventory of versioned files in a tree.
802
This describes which file_id is present at each point in the tree,
803
and possibly the SHA-1 or other information about the file.
804
Entries can be looked up either by path or by file_id.
806
The inventory represents a typical unix file tree, with
807
directories containing files and subdirectories. We never store
808
the full path to a file, because renaming a directory implicitly
809
moves all of its contents. This class internally maintains a
810
lookup tree that allows the children under a directory to be
813
InventoryEntry objects must not be modified after they are
814
inserted, other than through the Inventory API.
816
>>> inv = Inventory()
817
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
818
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT')
819
>>> inv['123-123'].name
822
May be treated as an iterator or set to look up file ids:
824
>>> bool(inv.path2id('hello.c'))
829
May also look up by name:
831
>>> [x[0] for x in inv.iter_entries()]
833
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
834
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
835
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT-12345678-12345678')
837
def __init__(self, root_id=ROOT_ID, revision_id=None):
838
"""Create or read an inventory.
840
If a working directory is specified, the inventory is read
841
from there. If the file is specified, read from that. If not,
842
the inventory is created empty.
844
The inventory is created with a default root directory, with
847
# We are letting Branch.create() create a unique inventory
848
# root id. Rather than generating a random one here.
850
# root_id = bzrlib.branch.gen_file_id('TREE_ROOT')
851
self.root = RootEntry(root_id)
852
self.revision_id = revision_id
853
self._byid = {self.root.file_id: self.root}
856
# TODO: jam 20051218 Should copy also copy the revision_id?
857
other = Inventory(self.root.file_id)
858
# copy recursively so we know directories will be added before
859
# their children. There are more efficient ways than this...
860
for path, entry in self.iter_entries():
861
if entry == self.root:
863
other.add(entry.copy())
867
return iter(self._byid)
870
"""Returns number of entries."""
871
return len(self._byid)
873
def iter_entries(self, from_dir=None):
874
"""Return (path, entry) pairs, in order by name."""
878
elif isinstance(from_dir, basestring):
879
from_dir = self._byid[from_dir]
881
# unrolling the recursive called changed the time from
882
# 440ms/663ms (inline/total) to 116ms/116ms
883
children = from_dir.children.items()
885
children = collections.deque(children)
886
stack = [(u'', children)]
888
from_dir_relpath, children = stack[-1]
891
name, ie = children.popleft()
893
# we know that from_dir_relpath never ends in a slash
894
# and 'f' doesn't begin with one, we can do a string op, rather
895
# than the checks of pathjoin(), though this means that all paths
897
path = from_dir_relpath + '/' + name
901
if ie.kind != 'directory':
904
# But do this child first
905
new_children = ie.children.items()
907
new_children = collections.deque(new_children)
908
stack.append((path, new_children))
909
# Break out of inner loop, so that we start outer loop with child
912
# if we finished all children, pop it off the stack
915
def iter_entries_by_dir(self, from_dir=None):
916
"""Iterate over the entries in a directory first order.
918
This returns all entries for a directory before returning
919
the entries for children of a directory. This is not
920
lexicographically sorted order, and is a hybrid between
921
depth-first and breadth-first.
923
:return: This yields (path, entry) pairs
925
# TODO? Perhaps this should return the from_dir so that the root is
926
# yielded? or maybe an option?
930
elif isinstance(from_dir, basestring):
931
from_dir = self._byid[from_dir]
933
stack = [(u'', from_dir)]
935
cur_relpath, cur_dir = stack.pop()
938
for child_name, child_ie in sorted(cur_dir.children.iteritems()):
940
child_relpath = cur_relpath + child_name
942
yield child_relpath, child_ie
944
if child_ie.kind == 'directory':
945
child_dirs.append((child_relpath+'/', child_ie))
946
stack.extend(reversed(child_dirs))
949
"""Return list of (path, ie) for all entries except the root.
951
This may be faster than iter_entries.
954
def descend(dir_ie, dir_path):
955
kids = dir_ie.children.items()
957
for name, ie in kids:
958
child_path = pathjoin(dir_path, name)
959
accum.append((child_path, ie))
960
if ie.kind == 'directory':
961
descend(ie, child_path)
963
descend(self.root, u'')
966
def directories(self):
967
"""Return (path, entry) pairs for all directories, including the root.
970
def descend(parent_ie, parent_path):
971
accum.append((parent_path, parent_ie))
973
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
976
for name, child_ie in kids:
977
child_path = pathjoin(parent_path, name)
978
descend(child_ie, child_path)
979
descend(self.root, u'')
982
def __contains__(self, file_id):
983
"""True if this entry contains a file with given id.
985
>>> inv = Inventory()
986
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
987
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT')
993
return file_id in self._byid
995
def __getitem__(self, file_id):
996
"""Return the entry for given file_id.
998
>>> inv = Inventory()
999
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
1000
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT')
1001
>>> inv['123123'].name
1005
return self._byid[file_id]
1008
raise BzrError("can't look up file_id None")
1010
raise BzrError("file_id {%s} not in inventory" % file_id)
1012
def get_file_kind(self, file_id):
1013
return self._byid[file_id].kind
1015
def get_child(self, parent_id, filename):
1016
return self[parent_id].children.get(filename)
1018
def add(self, entry):
1019
"""Add entry to inventory.
1021
To add a file to a branch ready to be committed, use Branch.add,
1024
Returns the new entry object.
1026
if entry.file_id in self._byid:
1027
raise BzrError("inventory already contains entry with id {%s}" % entry.file_id)
1029
if entry.parent_id == ROOT_ID or entry.parent_id is None:
1030
entry.parent_id = self.root.file_id
1033
parent = self._byid[entry.parent_id]
1035
raise BzrError("parent_id {%s} not in inventory" % entry.parent_id)
1037
if entry.name in parent.children:
1038
raise BzrError("%s is already versioned" %
1039
pathjoin(self.id2path(parent.file_id), entry.name))
1041
self._byid[entry.file_id] = entry
1042
parent.children[entry.name] = entry
1045
def add_path(self, relpath, kind, file_id=None, parent_id=None):
1046
"""Add entry from a path.
1048
The immediate parent must already be versioned.
1050
Returns the new entry object."""
1052
parts = bzrlib.osutils.splitpath(relpath)
1056
file_id = bzrlib.workingtree.gen_root_id()
1057
self.root = RootEntry(file_id)
1058
self._byid = {self.root.file_id: self.root}
1061
parent_path = parts[:-1]
1062
parent_id = self.path2id(parent_path)
1063
if parent_id is None:
1064
raise NotVersionedError(path=parent_path)
1065
ie = make_entry(kind, parts[-1], parent_id, file_id)
1068
def __delitem__(self, file_id):
1069
"""Remove entry by id.
1071
>>> inv = Inventory()
1072
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1073
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT')
1082
assert ie.parent_id is None or \
1083
self[ie.parent_id].children[ie.name] == ie
1085
del self._byid[file_id]
1086
if ie.parent_id is not None:
1087
del self[ie.parent_id].children[ie.name]
1089
def __eq__(self, other):
1090
"""Compare two sets by comparing their contents.
1092
>>> i1 = Inventory()
1093
>>> i2 = Inventory()
1096
>>> i1.add(InventoryFile('123', 'foo', ROOT_ID))
1097
InventoryFile('123', 'foo', parent_id='TREE_ROOT')
1100
>>> i2.add(InventoryFile('123', 'foo', ROOT_ID))
1101
InventoryFile('123', 'foo', parent_id='TREE_ROOT')
1105
if not isinstance(other, Inventory):
1106
return NotImplemented
1108
if len(self._byid) != len(other._byid):
1109
# shortcut: obviously not the same
1112
return self._byid == other._byid
1114
def __ne__(self, other):
1115
return not self.__eq__(other)
1118
raise ValueError('not hashable')
1120
def _iter_file_id_parents(self, file_id):
1121
"""Yield the parents of file_id up to the root."""
1122
while file_id is not None:
1124
ie = self._byid[file_id]
1126
raise BzrError("file_id {%s} not found in inventory" % file_id)
1128
file_id = ie.parent_id
1130
def get_idpath(self, file_id):
1131
"""Return a list of file_ids for the path to an entry.
1133
The list contains one element for each directory followed by
1134
the id of the file itself. So the length of the returned list
1135
is equal to the depth of the file in the tree, counting the
1136
root directory as depth 1.
1139
for parent in self._iter_file_id_parents(file_id):
1140
p.insert(0, parent.file_id)
1143
def id2path(self, file_id):
1144
"""Return as a string the path to file_id.
1147
>>> e = i.add(InventoryDirectory('src-id', 'src', ROOT_ID))
1148
>>> e = i.add(InventoryFile('foo-id', 'foo.c', parent_id='src-id'))
1149
>>> print i.id2path('foo-id')
1152
# get all names, skipping root
1153
return '/'.join(reversed(
1154
[parent.name for parent in
1155
self._iter_file_id_parents(file_id)][:-1]))
1157
def path2id(self, name):
1158
"""Walk down through directories to return entry of last component.
1160
names may be either a list of path components, or a single
1161
string, in which case it is automatically split.
1163
This returns the entry of the last component in the path,
1164
which may be either a file or a directory.
1166
Returns None IFF the path is not found.
1168
if isinstance(name, types.StringTypes):
1169
name = splitpath(name)
1171
# mutter("lookup path %r" % name)
1176
cie = parent.children[f]
1177
assert cie.name == f
1178
assert cie.parent_id == parent.file_id
1181
# or raise an error?
1184
return parent.file_id
1186
def has_filename(self, names):
1187
return bool(self.path2id(names))
1189
def has_id(self, file_id):
1190
return self._byid.has_key(file_id)
1192
def rename(self, file_id, new_parent_id, new_name):
1193
"""Move a file within the inventory.
1195
This can change either the name, or the parent, or both.
1197
This does not move the working file."""
1198
if not is_valid_name(new_name):
1199
raise BzrError("not an acceptable filename: %r" % new_name)
1201
new_parent = self._byid[new_parent_id]
1202
if new_name in new_parent.children:
1203
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
1205
new_parent_idpath = self.get_idpath(new_parent_id)
1206
if file_id in new_parent_idpath:
1207
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
1208
% (self.id2path(file_id), self.id2path(new_parent_id)))
1210
file_ie = self._byid[file_id]
1211
old_parent = self._byid[file_ie.parent_id]
1213
# TODO: Don't leave things messed up if this fails
1215
del old_parent.children[file_ie.name]
1216
new_parent.children[new_name] = file_ie
1218
file_ie.name = new_name
1219
file_ie.parent_id = new_parent_id
1222
def make_entry(kind, name, parent_id, file_id=None):
1223
"""Create an inventory entry.
1225
:param kind: the type of inventory entry to create.
1226
:param name: the basename of the entry.
1227
:param parent_id: the parent_id of the entry.
1228
:param file_id: the file_id to use. if None, one will be created.
1231
file_id = bzrlib.workingtree.gen_file_id(name)
1232
if kind == 'directory':
1233
return InventoryDirectory(file_id, name, parent_id)
1234
elif kind == 'file':
1235
return InventoryFile(file_id, name, parent_id)
1236
elif kind == 'symlink':
1237
return InventoryLink(file_id, name, parent_id)
1239
raise BzrError("unknown kind %r" % kind)
1245
def is_valid_name(name):
1247
if _NAME_RE is None:
1248
_NAME_RE = re.compile(r'^[^/\\]+$')
1250
return bool(_NAME_RE.match(name))
added
removed
bzrlib/inventory.py
