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'
126
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
127
'text_id', 'parent_id', 'children', 'executable',
130
def detect_changes(self, old_entry):
131
"""Return a (text_modified, meta_modified) from this to old_entry.
133
_read_tree_state must have been called on self and old_entry prior to
134
calling detect_changes.
138
def diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
139
output_to, reverse=False):
140
"""Perform a diff from this to to_entry.
142
text_diff will be used for textual difference calculation.
143
This is a template method, override _diff in child classes.
145
self._read_tree_state(tree.id2path(self.file_id), tree)
147
# cannot diff from one kind to another - you must do a removal
148
# and an addif they do not match.
149
assert self.kind == to_entry.kind
150
to_entry._read_tree_state(to_tree.id2path(to_entry.file_id),
152
self._diff(text_diff, from_label, tree, to_label, to_entry, to_tree,
155
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
156
output_to, reverse=False):
157
"""Perform a diff between two entries of the same kind."""
159
def find_previous_heads(self, previous_inventories,
160
versioned_file_store,
163
"""Return the revisions and entries that directly preceed this.
165
Returned as a map from revision to inventory entry.
167
This is a map containing the file revisions in all parents
168
for which the file exists, and its revision is not a parent of
169
any other. If the file is new, the set will be empty.
171
:param versioned_file_store: A store where ancestry data on this
172
file id can be queried.
173
:param transaction: The transaction that queries to the versioned
174
file store should be completed under.
175
:param entry_vf: The entry versioned file, if its already available.
177
def get_ancestors(weave, entry):
178
return set(weave.get_ancestry(entry.revision))
179
# revision:ie mapping for each ie found in previous_inventories.
181
# revision:ie mapping with one revision for each head.
183
# revision: ancestor list for each head
185
# identify candidate head revision ids.
186
for inv in previous_inventories:
187
if self.file_id in inv:
188
ie = inv[self.file_id]
189
assert ie.file_id == self.file_id
190
if ie.revision in candidates:
191
# same revision value in two different inventories:
192
# correct possible inconsistencies:
193
# * there was a bug in revision updates with 'x' bit
196
if candidates[ie.revision].executable != ie.executable:
197
candidates[ie.revision].executable = False
198
ie.executable = False
199
except AttributeError:
201
# must now be the same.
202
assert candidates[ie.revision] == ie
204
# add this revision as a candidate.
205
candidates[ie.revision] = ie
207
# common case optimisation
208
if len(candidates) == 1:
209
# if there is only one candidate revision found
210
# then we can opening the versioned file to access ancestry:
211
# there cannot be any ancestors to eliminate when there is
212
# only one revision available.
213
heads[ie.revision] = ie
216
# eliminate ancestors amongst the available candidates:
217
# heads are those that are not an ancestor of any other candidate
218
# - this provides convergence at a per-file level.
219
for ie in candidates.values():
220
# may be an ancestor of a known head:
221
already_present = 0 != len(
222
[head for head in heads
223
if ie.revision in head_ancestors[head]])
225
# an ancestor of an analyzed candidate.
227
# not an ancestor of a known head:
228
# load the versioned file for this file id if needed
230
entry_vf = versioned_file_store.get_weave_or_empty(
231
self.file_id, transaction)
232
ancestors = get_ancestors(entry_vf, ie)
233
# may knock something else out:
234
check_heads = list(heads.keys())
235
for head in check_heads:
236
if head in ancestors:
237
# this previously discovered 'head' is not
238
# really a head - its an ancestor of the newly
241
head_ancestors[ie.revision] = ancestors
242
heads[ie.revision] = ie
245
def get_tar_item(self, root, dp, now, tree):
246
"""Get a tarfile item and a file stream for its content."""
247
item = tarfile.TarInfo(pathjoin(root, dp))
248
# TODO: would be cool to actually set it to the timestamp of the
249
# revision it was last changed
251
fileobj = self._put_in_tar(item, tree)
255
"""Return true if the object this entry represents has textual data.
257
Note that textual data includes binary content.
259
Also note that all entries get weave files created for them.
260
This attribute is primarily used when upgrading from old trees that
261
did not have the weave index for all inventory entries.
265
def __init__(self, file_id, name, parent_id, text_id=None):
266
"""Create an InventoryEntry
268
The filename must be a single component, relative to the
269
parent directory; it cannot be a whole path or relative name.
271
>>> e = InventoryFile('123', 'hello.c', ROOT_ID)
276
>>> e = InventoryFile('123', 'src/hello.c', ROOT_ID)
277
Traceback (most recent call last):
278
InvalidEntryName: Invalid entry name: src/hello.c
280
assert isinstance(name, basestring), name
281
if '/' in name or '\\' in name:
282
raise InvalidEntryName(name=name)
283
self.executable = False
285
self.text_sha1 = None
286
self.text_size = None
287
self.file_id = file_id
289
self.text_id = text_id
290
self.parent_id = parent_id
291
self.symlink_target = None
293
def kind_character(self):
294
"""Return a short kind indicator useful for appending to names."""
295
raise BzrError('unknown kind %r' % self.kind)
297
known_kinds = ('file', 'directory', 'symlink', 'root_directory')
299
def _put_in_tar(self, item, tree):
300
"""populate item for stashing in a tar, and return the content stream.
302
If no content is available, return None.
304
raise BzrError("don't know how to export {%s} of kind %r" %
305
(self.file_id, self.kind))
307
def put_on_disk(self, dest, dp, tree):
308
"""Create a representation of self on disk in the prefix dest.
310
This is a template method - implement _put_on_disk in subclasses.
312
fullpath = pathjoin(dest, dp)
313
self._put_on_disk(fullpath, tree)
314
mutter(" export {%s} kind %s to %s", self.file_id,
317
def _put_on_disk(self, fullpath, tree):
318
"""Put this entry onto disk at fullpath, from tree tree."""
319
raise BzrError("don't know how to export {%s} of kind %r" % (self.file_id, self.kind))
321
def sorted_children(self):
322
l = self.children.items()
327
def versionable_kind(kind):
328
return kind in ('file', 'directory', 'symlink')
330
def check(self, checker, rev_id, inv, tree):
331
"""Check this inventory entry is intact.
333
This is a template method, override _check for kind specific
336
:param checker: Check object providing context for the checks;
337
can be used to find out what parts of the repository have already
339
:param rev_id: Revision id from which this InventoryEntry was loaded.
340
Not necessarily the last-changed revision for this file.
341
:param inv: Inventory from which the entry was loaded.
342
:param tree: RevisionTree for this entry.
344
if self.parent_id != None:
345
if not inv.has_id(self.parent_id):
346
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
347
% (self.parent_id, rev_id))
348
self._check(checker, rev_id, tree)
350
def _check(self, checker, rev_id, tree):
351
"""Check this inventory entry for kind specific errors."""
352
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
356
"""Clone this inventory entry."""
357
raise NotImplementedError
360
def describe_change(old_entry, new_entry):
361
"""Describe the change between old_entry and this.
363
This smells of being an InterInventoryEntry situation, but as its
364
the first one, we're making it a static method for now.
366
An entry with a different parent, or different name is considered
367
to be renamed. Reparenting is an internal detail.
368
Note that renaming the parent does not trigger a rename for the
371
# TODO: Perhaps return an object rather than just a string
372
if old_entry is new_entry:
373
# also the case of both being None
375
elif old_entry is None:
377
elif new_entry is None:
379
text_modified, meta_modified = new_entry.detect_changes(old_entry)
380
if text_modified or meta_modified:
384
# TODO 20060511 (mbp, rbc) factor out 'detect_rename' here.
385
if old_entry.parent_id != new_entry.parent_id:
387
elif old_entry.name != new_entry.name:
391
if renamed and not modified:
392
return InventoryEntry.RENAMED
393
if modified and not renamed:
395
if modified and renamed:
396
return InventoryEntry.MODIFIED_AND_RENAMED
400
return ("%s(%r, %r, parent_id=%r)"
401
% (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 overriden 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
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
def _check(self, checker, rev_id, tree):
524
"""See InventoryEntry._check"""
525
if self.text_sha1 != None or self.text_size != None or self.text_id != None:
526
raise BzrCheckError('directory {%s} has text in revision {%s}'
527
% (self.file_id, rev_id))
530
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
531
other.revision = self.revision
532
# note that children are *not* copied; they're pulled across when
536
def __init__(self, file_id, name, parent_id):
537
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
539
self.kind = 'directory'
541
def kind_character(self):
542
"""See InventoryEntry.kind_character."""
545
def _put_in_tar(self, item, tree):
546
"""See InventoryEntry._put_in_tar."""
547
item.type = tarfile.DIRTYPE
554
def _put_on_disk(self, fullpath, tree):
555
"""See InventoryEntry._put_on_disk."""
558
def _snapshot_text(self, file_parents, work_tree, commit_builder):
559
"""See InventoryEntry._snapshot_text."""
560
commit_builder.modified_directory(self.file_id, file_parents)
563
class InventoryFile(InventoryEntry):
564
"""A file in an inventory."""
566
def _check(self, checker, tree_revision_id, tree):
567
"""See InventoryEntry._check"""
568
t = (self.file_id, self.revision)
569
if t in checker.checked_texts:
570
prev_sha = checker.checked_texts[t]
571
if prev_sha != self.text_sha1:
572
raise BzrCheckError('mismatched sha1 on {%s} in {%s}' %
573
(self.file_id, tree_revision_id))
575
checker.repeated_text_cnt += 1
578
if self.file_id not in checker.checked_weaves:
579
mutter('check weave {%s}', self.file_id)
580
w = tree.get_weave(self.file_id)
581
# Not passing a progress bar, because it creates a new
582
# progress, which overwrites the current progress,
583
# and doesn't look nice
585
checker.checked_weaves[self.file_id] = True
587
w = tree.get_weave(self.file_id)
589
mutter('check version {%s} of {%s}', tree_revision_id, self.file_id)
590
checker.checked_text_cnt += 1
591
# We can't check the length, because Weave doesn't store that
592
# information, and the whole point of looking at the weave's
593
# sha1sum is that we don't have to extract the text.
594
if self.text_sha1 != w.get_sha1(self.revision):
595
raise BzrCheckError('text {%s} version {%s} wrong sha1'
596
% (self.file_id, self.revision))
597
checker.checked_texts[t] = self.text_sha1
600
other = InventoryFile(self.file_id, self.name, self.parent_id)
601
other.executable = self.executable
602
other.text_id = self.text_id
603
other.text_sha1 = self.text_sha1
604
other.text_size = self.text_size
605
other.revision = self.revision
608
def detect_changes(self, old_entry):
609
"""See InventoryEntry.detect_changes."""
610
assert self.text_sha1 != None
611
assert old_entry.text_sha1 != None
612
text_modified = (self.text_sha1 != old_entry.text_sha1)
613
meta_modified = (self.executable != old_entry.executable)
614
return text_modified, meta_modified
616
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
617
output_to, reverse=False):
618
"""See InventoryEntry._diff."""
620
from_text = tree.get_file(self.file_id).readlines()
622
to_text = to_tree.get_file(to_entry.file_id).readlines()
626
text_diff(from_label, from_text,
627
to_label, to_text, output_to)
629
text_diff(to_label, to_text,
630
from_label, from_text, output_to)
633
label_pair = (to_label, from_label)
635
label_pair = (from_label, to_label)
636
print >> output_to, "Binary files %s and %s differ" % label_pair
639
"""See InventoryEntry.has_text."""
642
def __init__(self, file_id, name, parent_id):
643
super(InventoryFile, self).__init__(file_id, name, parent_id)
646
def kind_character(self):
647
"""See InventoryEntry.kind_character."""
650
def _put_in_tar(self, item, tree):
651
"""See InventoryEntry._put_in_tar."""
652
item.type = tarfile.REGTYPE
653
fileobj = tree.get_file(self.file_id)
654
item.size = self.text_size
655
if tree.is_executable(self.file_id):
661
def _put_on_disk(self, fullpath, tree):
662
"""See InventoryEntry._put_on_disk."""
663
pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
664
if tree.is_executable(self.file_id):
665
os.chmod(fullpath, 0755)
667
def _read_tree_state(self, path, work_tree):
668
"""See InventoryEntry._read_tree_state."""
669
self.text_sha1 = work_tree.get_file_sha1(self.file_id, path=path)
670
# FIXME: 20050930 probe for the text size when getting sha1
671
# in _read_tree_state
672
self.executable = work_tree.is_executable(self.file_id, path=path)
674
def _forget_tree_state(self):
675
self.text_sha1 = None
676
self.executable = None
678
def _snapshot_text(self, file_parents, work_tree, commit_builder):
679
"""See InventoryEntry._snapshot_text."""
680
def get_content_byte_lines():
681
return work_tree.get_file(self.file_id).readlines()
682
self.text_sha1, self.text_size = commit_builder.modified_file_text(
683
self.file_id, file_parents, get_content_byte_lines, self.text_sha1, self.text_size)
685
def _unchanged(self, previous_ie):
686
"""See InventoryEntry._unchanged."""
687
compatible = super(InventoryFile, self)._unchanged(previous_ie)
688
if self.text_sha1 != previous_ie.text_sha1:
691
# FIXME: 20050930 probe for the text size when getting sha1
692
# in _read_tree_state
693
self.text_size = previous_ie.text_size
694
if self.executable != previous_ie.executable:
699
class InventoryLink(InventoryEntry):
700
"""A file in an inventory."""
702
__slots__ = ['symlink_target']
704
def _check(self, checker, rev_id, tree):
705
"""See InventoryEntry._check"""
706
if self.text_sha1 != None or self.text_size != None or self.text_id != None:
707
raise BzrCheckError('symlink {%s} has text in revision {%s}'
708
% (self.file_id, rev_id))
709
if self.symlink_target is None:
710
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
711
% (self.file_id, rev_id))
714
other = InventoryLink(self.file_id, self.name, self.parent_id)
715
other.symlink_target = self.symlink_target
716
other.revision = self.revision
719
def detect_changes(self, old_entry):
720
"""See InventoryEntry.detect_changes."""
721
# FIXME: which _modified field should we use ? RBC 20051003
722
text_modified = (self.symlink_target != old_entry.symlink_target)
724
mutter(" symlink target changed")
725
meta_modified = False
726
return text_modified, meta_modified
728
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
729
output_to, reverse=False):
730
"""See InventoryEntry._diff."""
731
from_text = self.symlink_target
732
if to_entry is not None:
733
to_text = to_entry.symlink_target
738
print >>output_to, '=== target changed %r => %r' % (from_text, to_text)
741
print >>output_to, '=== target was %r' % self.symlink_target
743
print >>output_to, '=== target is %r' % self.symlink_target
745
def __init__(self, file_id, name, parent_id):
746
super(InventoryLink, self).__init__(file_id, name, parent_id)
747
self.kind = 'symlink'
749
def kind_character(self):
750
"""See InventoryEntry.kind_character."""
753
def _put_in_tar(self, item, tree):
754
"""See InventoryEntry._put_in_tar."""
755
item.type = tarfile.SYMTYPE
759
item.linkname = self.symlink_target
762
def _put_on_disk(self, fullpath, tree):
763
"""See InventoryEntry._put_on_disk."""
765
os.symlink(self.symlink_target, fullpath)
767
raise BzrError("Failed to create symlink %r -> %r, error: %s" % (fullpath, self.symlink_target, e))
769
def _read_tree_state(self, path, work_tree):
770
"""See InventoryEntry._read_tree_state."""
771
self.symlink_target = work_tree.get_symlink_target(self.file_id)
773
def _forget_tree_state(self):
774
self.symlink_target = None
776
def _unchanged(self, previous_ie):
777
"""See InventoryEntry._unchanged."""
778
compatible = super(InventoryLink, self)._unchanged(previous_ie)
779
if self.symlink_target != previous_ie.symlink_target:
783
def _snapshot_text(self, file_parents, work_tree, commit_builder):
784
"""See InventoryEntry._snapshot_text."""
785
commit_builder.modified_link(
786
self.file_id, file_parents, self.symlink_target)
789
class Inventory(object):
790
"""Inventory of versioned files in a tree.
792
This describes which file_id is present at each point in the tree,
793
and possibly the SHA-1 or other information about the file.
794
Entries can be looked up either by path or by file_id.
796
The inventory represents a typical unix file tree, with
797
directories containing files and subdirectories. We never store
798
the full path to a file, because renaming a directory implicitly
799
moves all of its contents. This class internally maintains a
800
lookup tree that allows the children under a directory to be
803
InventoryEntry objects must not be modified after they are
804
inserted, other than through the Inventory API.
806
>>> inv = Inventory()
807
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
808
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT')
809
>>> inv['123-123'].name
812
May be treated as an iterator or set to look up file ids:
814
>>> bool(inv.path2id('hello.c'))
819
May also look up by name:
821
>>> [x[0] for x in inv.iter_entries()]
823
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
824
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
825
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT-12345678-12345678')
827
def __init__(self, root_id=ROOT_ID, revision_id=None):
828
"""Create or read an inventory.
830
If a working directory is specified, the inventory is read
831
from there. If the file is specified, read from that. If not,
832
the inventory is created empty.
834
The inventory is created with a default root directory, with
837
# We are letting Branch.create() create a unique inventory
838
# root id. Rather than generating a random one here.
840
# root_id = bzrlib.branch.gen_file_id('TREE_ROOT')
841
self.root = RootEntry(root_id)
842
self.revision_id = revision_id
843
self._byid = {self.root.file_id: self.root}
846
# TODO: jam 20051218 Should copy also copy the revision_id?
847
other = Inventory(self.root.file_id)
848
# copy recursively so we know directories will be added before
849
# their children. There are more efficient ways than this...
850
for path, entry in self.iter_entries():
851
if entry == self.root:
853
other.add(entry.copy())
857
return iter(self._byid)
860
"""Returns number of entries."""
861
return len(self._byid)
863
def iter_entries(self, from_dir=None):
864
"""Return (path, entry) pairs, in order by name."""
868
elif isinstance(from_dir, basestring):
869
from_dir = self._byid[from_dir]
871
# unrolling the recursive called changed the time from
872
# 440ms/663ms (inline/total) to 116ms/116ms
873
children = from_dir.children.items()
875
children = collections.deque(children)
876
stack = [(u'', children)]
878
from_dir_relpath, children = stack[-1]
881
name, ie = children.popleft()
883
# we know that from_dir_relpath never ends in a slash
884
# and 'f' doesn't begin with one, we can do a string op, rather
885
# than the checks of pathjoin(), though this means that all paths
887
path = from_dir_relpath + '/' + name
891
if ie.kind != 'directory':
894
# But do this child first
895
new_children = ie.children.items()
897
new_children = collections.deque(new_children)
898
stack.append((path, new_children))
899
# Break out of inner loop, so that we start outer loop with child
902
# if we finished all children, pop it off the stack
905
def iter_entries_by_dir(self, from_dir=None):
906
"""Iterate over the entries in a directory first order.
908
This returns all entries for a directory before returning
909
the entries for children of a directory. This is not
910
lexicographically sorted order, and is a hybrid between
911
depth-first and breadth-first.
913
:return: This yields (path, entry) pairs
915
# TODO? Perhaps this should return the from_dir so that the root is
916
# yielded? or maybe an option?
920
elif isinstance(from_dir, basestring):
921
from_dir = self._byid[from_dir]
923
stack = [(u'', from_dir)]
925
cur_relpath, cur_dir = stack.pop()
928
for child_name, child_ie in sorted(cur_dir.children.iteritems()):
930
child_relpath = cur_relpath + child_name
932
yield child_relpath, child_ie
934
if child_ie.kind == 'directory':
935
child_dirs.append((child_relpath+'/', child_ie))
936
stack.extend(reversed(child_dirs))
939
"""Return list of (path, ie) for all entries except the root.
941
This may be faster than iter_entries.
944
def descend(dir_ie, dir_path):
945
kids = dir_ie.children.items()
947
for name, ie in kids:
948
child_path = pathjoin(dir_path, name)
949
accum.append((child_path, ie))
950
if ie.kind == 'directory':
951
descend(ie, child_path)
953
descend(self.root, u'')
956
def directories(self):
957
"""Return (path, entry) pairs for all directories, including the root.
960
def descend(parent_ie, parent_path):
961
accum.append((parent_path, parent_ie))
963
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
966
for name, child_ie in kids:
967
child_path = pathjoin(parent_path, name)
968
descend(child_ie, child_path)
969
descend(self.root, u'')
972
def __contains__(self, file_id):
973
"""True if this entry contains a file with given id.
975
>>> inv = Inventory()
976
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
977
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT')
983
return file_id in self._byid
985
def __getitem__(self, file_id):
986
"""Return the entry for given file_id.
988
>>> inv = Inventory()
989
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
990
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT')
991
>>> inv['123123'].name
995
return self._byid[file_id]
998
raise BzrError("can't look up file_id None")
1000
raise BzrError("file_id {%s} not in inventory" % file_id)
1002
def get_file_kind(self, file_id):
1003
return self._byid[file_id].kind
1005
def get_child(self, parent_id, filename):
1006
return self[parent_id].children.get(filename)
1008
def add(self, entry):
1009
"""Add entry to inventory.
1011
To add a file to a branch ready to be committed, use Branch.add,
1014
Returns the new entry object.
1016
if entry.file_id in self._byid:
1017
raise BzrError("inventory already contains entry with id {%s}" % entry.file_id)
1019
if entry.parent_id == ROOT_ID or entry.parent_id is None:
1020
entry.parent_id = self.root.file_id
1023
parent = self._byid[entry.parent_id]
1025
raise BzrError("parent_id {%s} not in inventory" % entry.parent_id)
1027
if parent.children.has_key(entry.name):
1028
raise BzrError("%s is already versioned" %
1029
pathjoin(self.id2path(parent.file_id), entry.name))
1031
self._byid[entry.file_id] = entry
1032
parent.children[entry.name] = entry
1035
def add_path(self, relpath, kind, file_id=None, parent_id=None):
1036
"""Add entry from a path.
1038
The immediate parent must already be versioned.
1040
Returns the new entry object."""
1042
parts = bzrlib.osutils.splitpath(relpath)
1046
file_id = bzrlib.workingtree.gen_root_id()
1047
self.root = RootEntry(file_id)
1048
self._byid = {self.root.file_id: self.root}
1051
parent_path = parts[:-1]
1052
parent_id = self.path2id(parent_path)
1053
if parent_id is None:
1054
raise NotVersionedError(path=parent_path)
1055
ie = make_entry(kind, parts[-1], parent_id, file_id)
1058
def __delitem__(self, file_id):
1059
"""Remove entry by id.
1061
>>> inv = Inventory()
1062
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1063
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT')
1072
assert ie.parent_id is None or \
1073
self[ie.parent_id].children[ie.name] == ie
1075
del self._byid[file_id]
1076
if ie.parent_id is not None:
1077
del self[ie.parent_id].children[ie.name]
1079
def __eq__(self, other):
1080
"""Compare two sets by comparing their contents.
1082
>>> i1 = Inventory()
1083
>>> i2 = Inventory()
1086
>>> i1.add(InventoryFile('123', 'foo', ROOT_ID))
1087
InventoryFile('123', 'foo', parent_id='TREE_ROOT')
1090
>>> i2.add(InventoryFile('123', 'foo', ROOT_ID))
1091
InventoryFile('123', 'foo', parent_id='TREE_ROOT')
1095
if not isinstance(other, Inventory):
1096
return NotImplemented
1098
if len(self._byid) != len(other._byid):
1099
# shortcut: obviously not the same
1102
return self._byid == other._byid
1104
def __ne__(self, other):
1105
return not self.__eq__(other)
1108
raise ValueError('not hashable')
1110
def _iter_file_id_parents(self, file_id):
1111
"""Yield the parents of file_id up to the root."""
1112
while file_id != None:
1114
ie = self._byid[file_id]
1116
raise BzrError("file_id {%s} not found in inventory" % file_id)
1118
file_id = ie.parent_id
1120
def get_idpath(self, file_id):
1121
"""Return a list of file_ids for the path to an entry.
1123
The list contains one element for each directory followed by
1124
the id of the file itself. So the length of the returned list
1125
is equal to the depth of the file in the tree, counting the
1126
root directory as depth 1.
1129
for parent in self._iter_file_id_parents(file_id):
1130
p.insert(0, parent.file_id)
1133
def id2path(self, file_id):
1134
"""Return as a string the path to file_id.
1137
>>> e = i.add(InventoryDirectory('src-id', 'src', ROOT_ID))
1138
>>> e = i.add(InventoryFile('foo-id', 'foo.c', parent_id='src-id'))
1139
>>> print i.id2path('foo-id')
1142
# get all names, skipping root
1143
return '/'.join(reversed(
1144
[parent.name for parent in
1145
self._iter_file_id_parents(file_id)][:-1]))
1147
def path2id(self, name):
1148
"""Walk down through directories to return entry of last component.
1150
names may be either a list of path components, or a single
1151
string, in which case it is automatically split.
1153
This returns the entry of the last component in the path,
1154
which may be either a file or a directory.
1156
Returns None IFF the path is not found.
1158
if isinstance(name, types.StringTypes):
1159
name = splitpath(name)
1161
# mutter("lookup path %r" % name)
1166
cie = parent.children[f]
1167
assert cie.name == f
1168
assert cie.parent_id == parent.file_id
1171
# or raise an error?
1174
return parent.file_id
1176
def has_filename(self, names):
1177
return bool(self.path2id(names))
1179
def has_id(self, file_id):
1180
return self._byid.has_key(file_id)
1182
def rename(self, file_id, new_parent_id, new_name):
1183
"""Move a file within the inventory.
1185
This can change either the name, or the parent, or both.
1187
This does not move the working file."""
1188
if not is_valid_name(new_name):
1189
raise BzrError("not an acceptable filename: %r" % new_name)
1191
new_parent = self._byid[new_parent_id]
1192
if new_name in new_parent.children:
1193
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
1195
new_parent_idpath = self.get_idpath(new_parent_id)
1196
if file_id in new_parent_idpath:
1197
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
1198
% (self.id2path(file_id), self.id2path(new_parent_id)))
1200
file_ie = self._byid[file_id]
1201
old_parent = self._byid[file_ie.parent_id]
1203
# TODO: Don't leave things messed up if this fails
1205
del old_parent.children[file_ie.name]
1206
new_parent.children[new_name] = file_ie
1208
file_ie.name = new_name
1209
file_ie.parent_id = new_parent_id
1212
def make_entry(kind, name, parent_id, file_id=None):
1213
"""Create an inventory entry.
1215
:param kind: the type of inventory entry to create.
1216
:param name: the basename of the entry.
1217
:param parent_id: the parent_id of the entry.
1218
:param file_id: the file_id to use. if None, one will be created.
1221
file_id = bzrlib.workingtree.gen_file_id(name)
1222
if kind == 'directory':
1223
return InventoryDirectory(file_id, name, parent_id)
1224
elif kind == 'file':
1225
return InventoryFile(file_id, name, parent_id)
1226
elif kind == 'symlink':
1227
return InventoryLink(file_id, name, parent_id)
1229
raise BzrError("unknown kind %r" % kind)
1235
def is_valid_name(name):
1237
if _NAME_RE is None:
1238
_NAME_RE = re.compile(r'^[^/\\]+$')
1240
return bool(_NAME_RE.match(name))
added
removed
bzrlib/inventory.py
