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', revision=None)
81
>>> i.add(InventoryFile('2323', 'hello.c', parent_id='123'))
82
InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None)
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', revision=None))
88
(True, InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None))
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', sha1=None, len=None)
95
>>> i.add(InventoryDirectory('2325', 'wibble', '123'))
96
InventoryDirectory('2325', 'wibble', parent_id='123', revision=None)
97
>>> i.path2id('src/wibble')
101
>>> i.add(InventoryFile('2326', 'wibble.c', '2325'))
102
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
104
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
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, revision=%r)"
397
% (self.__class__.__name__,
403
def snapshot(self, revision, path, previous_entries,
404
work_tree, commit_builder):
405
"""Make a snapshot of this entry which may or may not have changed.
407
This means that all its fields are populated, that it has its
408
text stored in the text store or weave.
410
mutter('new parents of %s are %r', path, previous_entries)
411
self._read_tree_state(path, work_tree)
412
# TODO: Where should we determine whether to reuse a
413
# previous revision id or create a new revision? 20060606
414
if len(previous_entries) == 1:
415
# cannot be unchanged unless there is only one parent file rev.
416
parent_ie = previous_entries.values()[0]
417
if self._unchanged(parent_ie):
418
mutter("found unchanged entry")
419
self.revision = parent_ie.revision
421
return self._snapshot_into_revision(revision, previous_entries,
422
work_tree, commit_builder)
424
def _snapshot_into_revision(self, revision, previous_entries, work_tree,
426
"""Record this revision unconditionally into a store.
428
The entry's last-changed revision property (`revision`) is updated to
429
that of the new revision.
431
:param revision: id of the new revision that is being recorded.
433
:returns: String description of the commit (e.g. "merged", "modified"), etc.
435
mutter('new revision {%s} for {%s}', revision, self.file_id)
436
self.revision = revision
437
self._snapshot_text(previous_entries, work_tree, commit_builder)
439
def _snapshot_text(self, file_parents, work_tree, commit_builder):
440
"""Record the 'text' of this entry, whatever form that takes.
442
This default implementation simply adds an empty text.
444
raise NotImplementedError(self._snapshot_text)
446
def __eq__(self, other):
447
if not isinstance(other, InventoryEntry):
448
return NotImplemented
450
return ((self.file_id == other.file_id)
451
and (self.name == other.name)
452
and (other.symlink_target == self.symlink_target)
453
and (self.text_sha1 == other.text_sha1)
454
and (self.text_size == other.text_size)
455
and (self.text_id == other.text_id)
456
and (self.parent_id == other.parent_id)
457
and (self.kind == other.kind)
458
and (self.revision == other.revision)
459
and (self.executable == other.executable)
462
def __ne__(self, other):
463
return not (self == other)
466
raise ValueError('not hashable')
468
def _unchanged(self, previous_ie):
469
"""Has this entry changed relative to previous_ie.
471
This method should be overridden in child classes.
474
# different inv parent
475
if previous_ie.parent_id != self.parent_id:
478
elif previous_ie.name != self.name:
482
def _read_tree_state(self, path, work_tree):
483
"""Populate fields in the inventory entry from the given tree.
485
Note that this should be modified to be a noop on virtual trees
486
as all entries created there are prepopulated.
488
# TODO: Rather than running this manually, we should check the
489
# working sha1 and other expensive properties when they're
490
# first requested, or preload them if they're already known
491
pass # nothing to do by default
493
def _forget_tree_state(self):
497
class RootEntry(InventoryEntry):
499
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
500
'text_id', 'parent_id', 'children', 'executable',
501
'revision', 'symlink_target']
503
def _check(self, checker, rev_id, tree):
504
"""See InventoryEntry._check"""
506
def __init__(self, file_id):
507
self.file_id = file_id
509
self.kind = 'root_directory'
510
self.parent_id = None
514
def __eq__(self, other):
515
if not isinstance(other, RootEntry):
516
return NotImplemented
518
return (self.file_id == other.file_id) \
519
and (self.children == other.children)
522
class InventoryDirectory(InventoryEntry):
523
"""A directory in an inventory."""
525
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
526
'text_id', 'parent_id', 'children', 'executable',
527
'revision', 'symlink_target']
529
def _check(self, checker, rev_id, tree):
530
"""See InventoryEntry._check"""
531
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
532
raise BzrCheckError('directory {%s} has text in revision {%s}'
533
% (self.file_id, rev_id))
536
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
537
other.revision = self.revision
538
# note that children are *not* copied; they're pulled across when
542
def __init__(self, file_id, name, parent_id):
543
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
545
self.kind = 'directory'
547
def kind_character(self):
548
"""See InventoryEntry.kind_character."""
551
def _put_in_tar(self, item, tree):
552
"""See InventoryEntry._put_in_tar."""
553
item.type = tarfile.DIRTYPE
560
def _put_on_disk(self, fullpath, tree):
561
"""See InventoryEntry._put_on_disk."""
564
def _snapshot_text(self, file_parents, work_tree, commit_builder):
565
"""See InventoryEntry._snapshot_text."""
566
commit_builder.modified_directory(self.file_id, file_parents)
569
class InventoryFile(InventoryEntry):
570
"""A file in an inventory."""
572
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
573
'text_id', 'parent_id', 'children', 'executable',
574
'revision', 'symlink_target']
576
def _check(self, checker, tree_revision_id, tree):
577
"""See InventoryEntry._check"""
578
t = (self.file_id, self.revision)
579
if t in checker.checked_texts:
580
prev_sha = checker.checked_texts[t]
581
if prev_sha != self.text_sha1:
582
raise BzrCheckError('mismatched sha1 on {%s} in {%s}' %
583
(self.file_id, tree_revision_id))
585
checker.repeated_text_cnt += 1
588
if self.file_id not in checker.checked_weaves:
589
mutter('check weave {%s}', self.file_id)
590
w = tree.get_weave(self.file_id)
591
# Not passing a progress bar, because it creates a new
592
# progress, which overwrites the current progress,
593
# and doesn't look nice
595
checker.checked_weaves[self.file_id] = True
597
w = tree.get_weave(self.file_id)
599
mutter('check version {%s} of {%s}', tree_revision_id, self.file_id)
600
checker.checked_text_cnt += 1
601
# We can't check the length, because Weave doesn't store that
602
# information, and the whole point of looking at the weave's
603
# sha1sum is that we don't have to extract the text.
604
if self.text_sha1 != w.get_sha1(self.revision):
605
raise BzrCheckError('text {%s} version {%s} wrong sha1'
606
% (self.file_id, self.revision))
607
checker.checked_texts[t] = self.text_sha1
610
other = InventoryFile(self.file_id, self.name, self.parent_id)
611
other.executable = self.executable
612
other.text_id = self.text_id
613
other.text_sha1 = self.text_sha1
614
other.text_size = self.text_size
615
other.revision = self.revision
618
def detect_changes(self, old_entry):
619
"""See InventoryEntry.detect_changes."""
620
assert self.text_sha1 is not None
621
assert old_entry.text_sha1 is not None
622
text_modified = (self.text_sha1 != old_entry.text_sha1)
623
meta_modified = (self.executable != old_entry.executable)
624
return text_modified, meta_modified
626
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
627
output_to, reverse=False):
628
"""See InventoryEntry._diff."""
630
from_text = tree.get_file(self.file_id).readlines()
632
to_text = to_tree.get_file(to_entry.file_id).readlines()
636
text_diff(from_label, from_text,
637
to_label, to_text, output_to)
639
text_diff(to_label, to_text,
640
from_label, from_text, output_to)
643
label_pair = (to_label, from_label)
645
label_pair = (from_label, to_label)
646
print >> output_to, "Binary files %s and %s differ" % label_pair
649
"""See InventoryEntry.has_text."""
652
def __init__(self, file_id, name, parent_id):
653
super(InventoryFile, self).__init__(file_id, name, parent_id)
656
def kind_character(self):
657
"""See InventoryEntry.kind_character."""
660
def _put_in_tar(self, item, tree):
661
"""See InventoryEntry._put_in_tar."""
662
item.type = tarfile.REGTYPE
663
fileobj = tree.get_file(self.file_id)
664
item.size = self.text_size
665
if tree.is_executable(self.file_id):
671
def _put_on_disk(self, fullpath, tree):
672
"""See InventoryEntry._put_on_disk."""
673
pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
674
if tree.is_executable(self.file_id):
675
os.chmod(fullpath, 0755)
677
def _read_tree_state(self, path, work_tree):
678
"""See InventoryEntry._read_tree_state."""
679
self.text_sha1 = work_tree.get_file_sha1(self.file_id, path=path)
680
# FIXME: 20050930 probe for the text size when getting sha1
681
# in _read_tree_state
682
self.executable = work_tree.is_executable(self.file_id, path=path)
685
return ("%s(%r, %r, parent_id=%r, sha1=%r, len=%s)"
686
% (self.__class__.__name__,
693
def _forget_tree_state(self):
694
self.text_sha1 = None
696
def _snapshot_text(self, file_parents, work_tree, commit_builder):
697
"""See InventoryEntry._snapshot_text."""
698
def get_content_byte_lines():
699
return work_tree.get_file(self.file_id).readlines()
700
self.text_sha1, self.text_size = commit_builder.modified_file_text(
701
self.file_id, file_parents, get_content_byte_lines, self.text_sha1, self.text_size)
703
def _unchanged(self, previous_ie):
704
"""See InventoryEntry._unchanged."""
705
compatible = super(InventoryFile, self)._unchanged(previous_ie)
706
if self.text_sha1 != previous_ie.text_sha1:
709
# FIXME: 20050930 probe for the text size when getting sha1
710
# in _read_tree_state
711
self.text_size = previous_ie.text_size
712
if self.executable != previous_ie.executable:
717
class InventoryLink(InventoryEntry):
718
"""A file in an inventory."""
720
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
721
'text_id', 'parent_id', 'children', 'executable',
722
'revision', 'symlink_target']
724
def _check(self, checker, rev_id, tree):
725
"""See InventoryEntry._check"""
726
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
727
raise BzrCheckError('symlink {%s} has text in revision {%s}'
728
% (self.file_id, rev_id))
729
if self.symlink_target is None:
730
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
731
% (self.file_id, rev_id))
734
other = InventoryLink(self.file_id, self.name, self.parent_id)
735
other.symlink_target = self.symlink_target
736
other.revision = self.revision
739
def detect_changes(self, old_entry):
740
"""See InventoryEntry.detect_changes."""
741
# FIXME: which _modified field should we use ? RBC 20051003
742
text_modified = (self.symlink_target != old_entry.symlink_target)
744
mutter(" symlink target changed")
745
meta_modified = False
746
return text_modified, meta_modified
748
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
749
output_to, reverse=False):
750
"""See InventoryEntry._diff."""
751
from_text = self.symlink_target
752
if to_entry is not None:
753
to_text = to_entry.symlink_target
758
print >>output_to, '=== target changed %r => %r' % (from_text, to_text)
761
print >>output_to, '=== target was %r' % self.symlink_target
763
print >>output_to, '=== target is %r' % self.symlink_target
765
def __init__(self, file_id, name, parent_id):
766
super(InventoryLink, self).__init__(file_id, name, parent_id)
767
self.kind = 'symlink'
769
def kind_character(self):
770
"""See InventoryEntry.kind_character."""
773
def _put_in_tar(self, item, tree):
774
"""See InventoryEntry._put_in_tar."""
775
item.type = tarfile.SYMTYPE
779
item.linkname = self.symlink_target
782
def _put_on_disk(self, fullpath, tree):
783
"""See InventoryEntry._put_on_disk."""
785
os.symlink(self.symlink_target, fullpath)
787
raise BzrError("Failed to create symlink %r -> %r, error: %s" % (fullpath, self.symlink_target, e))
789
def _read_tree_state(self, path, work_tree):
790
"""See InventoryEntry._read_tree_state."""
791
self.symlink_target = work_tree.get_symlink_target(self.file_id)
793
def _forget_tree_state(self):
794
self.symlink_target = None
796
def _unchanged(self, previous_ie):
797
"""See InventoryEntry._unchanged."""
798
compatible = super(InventoryLink, self)._unchanged(previous_ie)
799
if self.symlink_target != previous_ie.symlink_target:
803
def _snapshot_text(self, file_parents, work_tree, commit_builder):
804
"""See InventoryEntry._snapshot_text."""
805
commit_builder.modified_link(
806
self.file_id, file_parents, self.symlink_target)
809
class Inventory(object):
810
"""Inventory of versioned files in a tree.
812
This describes which file_id is present at each point in the tree,
813
and possibly the SHA-1 or other information about the file.
814
Entries can be looked up either by path or by file_id.
816
The inventory represents a typical unix file tree, with
817
directories containing files and subdirectories. We never store
818
the full path to a file, because renaming a directory implicitly
819
moves all of its contents. This class internally maintains a
820
lookup tree that allows the children under a directory to be
823
InventoryEntry objects must not be modified after they are
824
inserted, other than through the Inventory API.
826
>>> inv = Inventory()
827
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
828
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
829
>>> inv['123-123'].name
832
May be treated as an iterator or set to look up file ids:
834
>>> bool(inv.path2id('hello.c'))
839
May also look up by name:
841
>>> [x[0] for x in inv.iter_entries()]
843
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
844
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
845
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT-12345678-12345678', sha1=None, len=None)
847
def __init__(self, root_id=ROOT_ID, revision_id=None):
848
"""Create or read an inventory.
850
If a working directory is specified, the inventory is read
851
from there. If the file is specified, read from that. If not,
852
the inventory is created empty.
854
The inventory is created with a default root directory, with
857
# We are letting Branch.create() create a unique inventory
858
# root id. Rather than generating a random one here.
860
# root_id = bzrlib.branch.gen_file_id('TREE_ROOT')
861
self.root = RootEntry(root_id)
862
# FIXME: this isn't ever used, changing it to self.revision may break
863
# things. TODO make everything use self.revision_id
864
self.revision_id = revision_id
865
self._byid = {self.root.file_id: self.root}
868
# TODO: jam 20051218 Should copy also copy the revision_id?
869
other = Inventory(self.root.file_id)
870
# copy recursively so we know directories will be added before
871
# their children. There are more efficient ways than this...
872
for path, entry in self.iter_entries():
873
if entry == self.root:
875
other.add(entry.copy())
879
return iter(self._byid)
882
"""Returns number of entries."""
883
return len(self._byid)
885
def iter_entries(self, from_dir=None):
886
"""Return (path, entry) pairs, in order by name."""
890
elif isinstance(from_dir, basestring):
891
from_dir = self._byid[from_dir]
893
# unrolling the recursive called changed the time from
894
# 440ms/663ms (inline/total) to 116ms/116ms
895
children = from_dir.children.items()
897
children = collections.deque(children)
898
stack = [(u'', children)]
900
from_dir_relpath, children = stack[-1]
903
name, ie = children.popleft()
905
# we know that from_dir_relpath never ends in a slash
906
# and 'f' doesn't begin with one, we can do a string op, rather
907
# than the checks of pathjoin(), though this means that all paths
909
path = from_dir_relpath + '/' + name
913
if ie.kind != 'directory':
916
# But do this child first
917
new_children = ie.children.items()
919
new_children = collections.deque(new_children)
920
stack.append((path, new_children))
921
# Break out of inner loop, so that we start outer loop with child
924
# if we finished all children, pop it off the stack
927
def iter_entries_by_dir(self, from_dir=None):
928
"""Iterate over the entries in a directory first order.
930
This returns all entries for a directory before returning
931
the entries for children of a directory. This is not
932
lexicographically sorted order, and is a hybrid between
933
depth-first and breadth-first.
935
:return: This yields (path, entry) pairs
937
# TODO? Perhaps this should return the from_dir so that the root is
938
# yielded? or maybe an option?
942
elif isinstance(from_dir, basestring):
943
from_dir = self._byid[from_dir]
945
stack = [(u'', from_dir)]
947
cur_relpath, cur_dir = stack.pop()
950
for child_name, child_ie in sorted(cur_dir.children.iteritems()):
952
child_relpath = cur_relpath + child_name
954
yield child_relpath, child_ie
956
if child_ie.kind == 'directory':
957
child_dirs.append((child_relpath+'/', child_ie))
958
stack.extend(reversed(child_dirs))
961
"""Return list of (path, ie) for all entries except the root.
963
This may be faster than iter_entries.
966
def descend(dir_ie, dir_path):
967
kids = dir_ie.children.items()
969
for name, ie in kids:
970
child_path = pathjoin(dir_path, name)
971
accum.append((child_path, ie))
972
if ie.kind == 'directory':
973
descend(ie, child_path)
975
descend(self.root, u'')
978
def directories(self):
979
"""Return (path, entry) pairs for all directories, including the root.
982
def descend(parent_ie, parent_path):
983
accum.append((parent_path, parent_ie))
985
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
988
for name, child_ie in kids:
989
child_path = pathjoin(parent_path, name)
990
descend(child_ie, child_path)
991
descend(self.root, u'')
994
def __contains__(self, file_id):
995
"""True if this entry contains a file with given id.
997
>>> inv = Inventory()
998
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
999
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
1005
return file_id in self._byid
1007
def __getitem__(self, file_id):
1008
"""Return the entry for given file_id.
1010
>>> inv = Inventory()
1011
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
1012
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
1013
>>> inv['123123'].name
1017
return self._byid[file_id]
1020
raise BzrError("can't look up file_id None")
1022
raise BzrError("file_id {%s} not in inventory" % file_id)
1024
def get_file_kind(self, file_id):
1025
return self._byid[file_id].kind
1027
def get_child(self, parent_id, filename):
1028
return self[parent_id].children.get(filename)
1030
def add(self, entry):
1031
"""Add entry to inventory.
1033
To add a file to a branch ready to be committed, use Branch.add,
1036
Returns the new entry object.
1038
if entry.file_id in self._byid:
1039
raise BzrError("inventory already contains entry with id {%s}" % entry.file_id)
1041
if entry.parent_id == ROOT_ID or entry.parent_id is None:
1042
entry.parent_id = self.root.file_id
1045
parent = self._byid[entry.parent_id]
1047
raise BzrError("parent_id {%s} not in inventory" % entry.parent_id)
1049
if entry.name in parent.children:
1050
raise BzrError("%s is already versioned" %
1051
pathjoin(self.id2path(parent.file_id), entry.name))
1053
self._byid[entry.file_id] = entry
1054
parent.children[entry.name] = entry
1057
def add_path(self, relpath, kind, file_id=None, parent_id=None):
1058
"""Add entry from a path.
1060
The immediate parent must already be versioned.
1062
Returns the new entry object."""
1064
parts = bzrlib.osutils.splitpath(relpath)
1068
file_id = bzrlib.workingtree.gen_root_id()
1069
self.root = RootEntry(file_id)
1070
self._byid = {self.root.file_id: self.root}
1073
parent_path = parts[:-1]
1074
parent_id = self.path2id(parent_path)
1075
if parent_id is None:
1076
raise NotVersionedError(path=parent_path)
1077
ie = make_entry(kind, parts[-1], parent_id, file_id)
1080
def __delitem__(self, file_id):
1081
"""Remove entry by id.
1083
>>> inv = Inventory()
1084
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1085
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
1094
assert ie.parent_id is None or \
1095
self[ie.parent_id].children[ie.name] == ie
1097
del self._byid[file_id]
1098
if ie.parent_id is not None:
1099
del self[ie.parent_id].children[ie.name]
1101
def __eq__(self, other):
1102
"""Compare two sets by comparing their contents.
1104
>>> i1 = Inventory()
1105
>>> i2 = Inventory()
1108
>>> i1.add(InventoryFile('123', 'foo', ROOT_ID))
1109
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
1112
>>> i2.add(InventoryFile('123', 'foo', ROOT_ID))
1113
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
1117
if not isinstance(other, Inventory):
1118
return NotImplemented
1120
return self._byid == other._byid
1122
def __ne__(self, other):
1123
return not self.__eq__(other)
1126
raise ValueError('not hashable')
1128
def _iter_file_id_parents(self, file_id):
1129
"""Yield the parents of file_id up to the root."""
1130
while file_id is not None:
1132
ie = self._byid[file_id]
1134
raise BzrError("file_id {%s} not found in inventory" % file_id)
1136
file_id = ie.parent_id
1138
def get_idpath(self, file_id):
1139
"""Return a list of file_ids for the path to an entry.
1141
The list contains one element for each directory followed by
1142
the id of the file itself. So the length of the returned list
1143
is equal to the depth of the file in the tree, counting the
1144
root directory as depth 1.
1147
for parent in self._iter_file_id_parents(file_id):
1148
p.insert(0, parent.file_id)
1151
def id2path(self, file_id):
1152
"""Return as a string the path to file_id.
1155
>>> e = i.add(InventoryDirectory('src-id', 'src', ROOT_ID))
1156
>>> e = i.add(InventoryFile('foo-id', 'foo.c', parent_id='src-id'))
1157
>>> print i.id2path('foo-id')
1160
# get all names, skipping root
1161
return '/'.join(reversed(
1162
[parent.name for parent in
1163
self._iter_file_id_parents(file_id)][:-1]))
1165
def path2id(self, name):
1166
"""Walk down through directories to return entry of last component.
1168
names may be either a list of path components, or a single
1169
string, in which case it is automatically split.
1171
This returns the entry of the last component in the path,
1172
which may be either a file or a directory.
1174
Returns None IFF the path is not found.
1176
if isinstance(name, types.StringTypes):
1177
name = splitpath(name)
1179
# mutter("lookup path %r" % name)
1184
cie = parent.children[f]
1185
assert cie.name == f
1186
assert cie.parent_id == parent.file_id
1189
# or raise an error?
1192
return parent.file_id
1194
def has_filename(self, names):
1195
return bool(self.path2id(names))
1197
def has_id(self, file_id):
1198
return self._byid.has_key(file_id)
1200
def rename(self, file_id, new_parent_id, new_name):
1201
"""Move a file within the inventory.
1203
This can change either the name, or the parent, or both.
1205
This does not move the working file."""
1206
if not is_valid_name(new_name):
1207
raise BzrError("not an acceptable filename: %r" % new_name)
1209
new_parent = self._byid[new_parent_id]
1210
if new_name in new_parent.children:
1211
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
1213
new_parent_idpath = self.get_idpath(new_parent_id)
1214
if file_id in new_parent_idpath:
1215
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
1216
% (self.id2path(file_id), self.id2path(new_parent_id)))
1218
file_ie = self._byid[file_id]
1219
old_parent = self._byid[file_ie.parent_id]
1221
# TODO: Don't leave things messed up if this fails
1223
del old_parent.children[file_ie.name]
1224
new_parent.children[new_name] = file_ie
1226
file_ie.name = new_name
1227
file_ie.parent_id = new_parent_id
1230
def make_entry(kind, name, parent_id, file_id=None):
1231
"""Create an inventory entry.
1233
:param kind: the type of inventory entry to create.
1234
:param name: the basename of the entry.
1235
:param parent_id: the parent_id of the entry.
1236
:param file_id: the file_id to use. if None, one will be created.
1239
file_id = bzrlib.workingtree.gen_file_id(name)
1240
if kind == 'directory':
1241
return InventoryDirectory(file_id, name, parent_id)
1242
elif kind == 'file':
1243
return InventoryFile(file_id, name, parent_id)
1244
elif kind == 'symlink':
1245
return InventoryLink(file_id, name, parent_id)
1247
raise BzrError("unknown kind %r" % kind)
1253
def is_valid_name(name):
1255
if _NAME_RE is None:
1256
_NAME_RE = re.compile(r'^[^/\\]+$')
1258
return bool(_NAME_RE.match(name))
added
removed
bzrlib/inventory.py
