1
# Copyright (C) 2005, 2006 Canonical Ltd
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
# FIXME: This refactoring of the workingtree code doesn't seem to keep
18
# the WorkingTree's copy of the inventory in sync with the branch. The
19
# branch modifies its working inventory when it does a commit to make
20
# missing files permanently removed.
22
# TODO: Maybe also keep the full path of the entry, and the children?
23
# But those depend on its position within a particular inventory, and
24
# it would be nice not to need to hold the backpointer here.
26
# This should really be an id randomly assigned when the tree is
27
# created, but it's not for now.
39
from bzrlib import errors, osutils
40
from bzrlib.osutils import (pumpfile, quotefn, splitpath, joinpath,
41
pathjoin, sha_strings)
42
from bzrlib.errors import (NotVersionedError, InvalidEntryName,
43
BzrError, BzrCheckError, BinaryFile)
44
from bzrlib.trace import mutter
47
class InventoryEntry(object):
48
"""Description of a versioned file.
50
An InventoryEntry has the following fields, which are also
51
present in the XML inventory-entry element:
56
(within the parent directory)
59
file_id of the parent directory, or ROOT_ID
62
the revision_id in which this variation of this file was
66
Indicates that this file should be executable on systems
70
sha-1 of the text of the file
73
size in bytes of the text of the file
75
(reading a version 4 tree created a text_id field.)
80
>>> i.add(InventoryDirectory('123', 'src', ROOT_ID))
81
InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None)
82
>>> i.add(InventoryFile('2323', 'hello.c', parent_id='123'))
83
InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None)
84
>>> shouldbe = {0: 'src', 1: pathjoin('src','hello.c')}
85
>>> for ix, j in enumerate(i.iter_entries()):
86
... print (j[0] == shouldbe[ix], j[1])
88
(True, InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None))
89
(True, InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None))
90
>>> i.add(InventoryFile('2323', 'bye.c', '123'))
91
Traceback (most recent call last):
93
BzrError: inventory already contains entry with id {2323}
94
>>> i.add(InventoryFile('2324', 'bye.c', '123'))
95
InventoryFile('2324', 'bye.c', parent_id='123', sha1=None, len=None)
96
>>> i.add(InventoryDirectory('2325', 'wibble', '123'))
97
InventoryDirectory('2325', 'wibble', parent_id='123', revision=None)
98
>>> i.path2id('src/wibble')
102
>>> i.add(InventoryFile('2326', 'wibble.c', '2325'))
103
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
105
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
106
>>> for path, entry in i.iter_entries():
108
... assert i.path2id(path)
115
>>> i.id2path('2326')
116
'src/wibble/wibble.c'
119
# Constants returned by describe_change()
121
# TODO: These should probably move to some kind of FileChangeDescription
122
# class; that's like what's inside a TreeDelta but we want to be able to
123
# generate them just for one file at a time.
125
MODIFIED_AND_RENAMED = 'modified and renamed'
129
def detect_changes(self, old_entry):
130
"""Return a (text_modified, meta_modified) from this to old_entry.
132
_read_tree_state must have been called on self and old_entry prior to
133
calling detect_changes.
137
def diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
138
output_to, reverse=False):
139
"""Perform a diff from this to to_entry.
141
text_diff will be used for textual difference calculation.
142
This is a template method, override _diff in child classes.
144
self._read_tree_state(tree.id2path(self.file_id), tree)
146
# cannot diff from one kind to another - you must do a removal
147
# and an addif they do not match.
148
assert self.kind == to_entry.kind
149
to_entry._read_tree_state(to_tree.id2path(to_entry.file_id),
151
self._diff(text_diff, from_label, tree, to_label, to_entry, to_tree,
154
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
155
output_to, reverse=False):
156
"""Perform a diff between two entries of the same kind."""
158
def find_previous_heads(self, previous_inventories,
159
versioned_file_store,
162
"""Return the revisions and entries that directly precede this.
164
Returned as a map from revision to inventory entry.
166
This is a map containing the file revisions in all parents
167
for which the file exists, and its revision is not a parent of
168
any other. If the file is new, the set will be empty.
170
:param versioned_file_store: A store where ancestry data on this
171
file id can be queried.
172
:param transaction: The transaction that queries to the versioned
173
file store should be completed under.
174
:param entry_vf: The entry versioned file, if its already available.
176
def get_ancestors(weave, entry):
177
return set(weave.get_ancestry(entry.revision))
178
# revision:ie mapping for each ie found in previous_inventories.
180
# revision:ie mapping with one revision for each head.
182
# revision: ancestor list for each head
184
# identify candidate head revision ids.
185
for inv in previous_inventories:
186
if self.file_id in inv:
187
ie = inv[self.file_id]
188
assert ie.file_id == self.file_id
189
if ie.revision in candidates:
190
# same revision value in two different inventories:
191
# correct possible inconsistencies:
192
# * there was a bug in revision updates with 'x' bit
195
if candidates[ie.revision].executable != ie.executable:
196
candidates[ie.revision].executable = False
197
ie.executable = False
198
except AttributeError:
200
# must now be the same.
201
assert candidates[ie.revision] == ie
203
# add this revision as a candidate.
204
candidates[ie.revision] = ie
206
# common case optimisation
207
if len(candidates) == 1:
208
# if there is only one candidate revision found
209
# then we can opening the versioned file to access ancestry:
210
# there cannot be any ancestors to eliminate when there is
211
# only one revision available.
212
heads[ie.revision] = ie
215
# eliminate ancestors amongst the available candidates:
216
# heads are those that are not an ancestor of any other candidate
217
# - this provides convergence at a per-file level.
218
for ie in candidates.values():
219
# may be an ancestor of a known head:
220
already_present = 0 != len(
221
[head for head in heads
222
if ie.revision in head_ancestors[head]])
224
# an ancestor of an analyzed candidate.
226
# not an ancestor of a known head:
227
# load the versioned file for this file id if needed
229
entry_vf = versioned_file_store.get_weave_or_empty(
230
self.file_id, transaction)
231
ancestors = get_ancestors(entry_vf, ie)
232
# may knock something else out:
233
check_heads = list(heads.keys())
234
for head in check_heads:
235
if head in ancestors:
236
# this previously discovered 'head' is not
237
# really a head - its an ancestor of the newly
240
head_ancestors[ie.revision] = ancestors
241
heads[ie.revision] = ie
244
def get_tar_item(self, root, dp, now, tree):
245
"""Get a tarfile item and a file stream for its content."""
246
item = tarfile.TarInfo(pathjoin(root, dp))
247
# TODO: would be cool to actually set it to the timestamp of the
248
# revision it was last changed
250
fileobj = self._put_in_tar(item, tree)
254
"""Return true if the object this entry represents has textual data.
256
Note that textual data includes binary content.
258
Also note that all entries get weave files created for them.
259
This attribute is primarily used when upgrading from old trees that
260
did not have the weave index for all inventory entries.
264
def __init__(self, file_id, name, parent_id, text_id=None):
265
"""Create an InventoryEntry
267
The filename must be a single component, relative to the
268
parent directory; it cannot be a whole path or relative name.
270
>>> e = InventoryFile('123', 'hello.c', ROOT_ID)
275
>>> e = InventoryFile('123', 'src/hello.c', ROOT_ID)
276
Traceback (most recent call last):
277
InvalidEntryName: Invalid entry name: src/hello.c
279
assert isinstance(name, basestring), name
280
if '/' in name or '\\' in name:
281
raise InvalidEntryName(name=name)
282
self.executable = False
284
self.text_sha1 = None
285
self.text_size = None
286
self.file_id = file_id
288
self.text_id = text_id
289
self.parent_id = parent_id
290
self.symlink_target = None
292
def kind_character(self):
293
"""Return a short kind indicator useful for appending to names."""
294
raise BzrError('unknown kind %r' % self.kind)
296
known_kinds = ('file', 'directory', 'symlink', 'root_directory')
298
def _put_in_tar(self, item, tree):
299
"""populate item for stashing in a tar, and return the content stream.
301
If no content is available, return None.
303
raise BzrError("don't know how to export {%s} of kind %r" %
304
(self.file_id, self.kind))
306
def put_on_disk(self, dest, dp, tree):
307
"""Create a representation of self on disk in the prefix dest.
309
This is a template method - implement _put_on_disk in subclasses.
311
fullpath = pathjoin(dest, dp)
312
self._put_on_disk(fullpath, tree)
313
mutter(" export {%s} kind %s to %s", self.file_id,
316
def _put_on_disk(self, fullpath, tree):
317
"""Put this entry onto disk at fullpath, from tree tree."""
318
raise BzrError("don't know how to export {%s} of kind %r" % (self.file_id, self.kind))
320
def sorted_children(self):
321
return sorted(self.children.items())
324
def versionable_kind(kind):
325
return kind in ('file', 'directory', 'symlink')
327
def check(self, checker, rev_id, inv, tree):
328
"""Check this inventory entry is intact.
330
This is a template method, override _check for kind specific
333
:param checker: Check object providing context for the checks;
334
can be used to find out what parts of the repository have already
336
:param rev_id: Revision id from which this InventoryEntry was loaded.
337
Not necessarily the last-changed revision for this file.
338
:param inv: Inventory from which the entry was loaded.
339
:param tree: RevisionTree for this entry.
341
if self.parent_id is not None:
342
if not inv.has_id(self.parent_id):
343
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
344
% (self.parent_id, rev_id))
345
self._check(checker, rev_id, tree)
347
def _check(self, checker, rev_id, tree):
348
"""Check this inventory entry for kind specific errors."""
349
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
353
"""Clone this inventory entry."""
354
raise NotImplementedError
357
def describe_change(old_entry, new_entry):
358
"""Describe the change between old_entry and this.
360
This smells of being an InterInventoryEntry situation, but as its
361
the first one, we're making it a static method for now.
363
An entry with a different parent, or different name is considered
364
to be renamed. Reparenting is an internal detail.
365
Note that renaming the parent does not trigger a rename for the
368
# TODO: Perhaps return an object rather than just a string
369
if old_entry is new_entry:
370
# also the case of both being None
372
elif old_entry is None:
374
elif new_entry is None:
376
text_modified, meta_modified = new_entry.detect_changes(old_entry)
377
if text_modified or meta_modified:
381
# TODO 20060511 (mbp, rbc) factor out 'detect_rename' here.
382
if old_entry.parent_id != new_entry.parent_id:
384
elif old_entry.name != new_entry.name:
388
if renamed and not modified:
389
return InventoryEntry.RENAMED
390
if modified and not renamed:
392
if modified and renamed:
393
return InventoryEntry.MODIFIED_AND_RENAMED
397
return ("%s(%r, %r, parent_id=%r, revision=%r)"
398
% (self.__class__.__name__,
404
def snapshot(self, revision, path, previous_entries,
405
work_tree, commit_builder):
406
"""Make a snapshot of this entry which may or may not have changed.
408
This means that all its fields are populated, that it has its
409
text stored in the text store or weave.
411
mutter('new parents of %s are %r', path, previous_entries)
412
self._read_tree_state(path, work_tree)
413
# TODO: Where should we determine whether to reuse a
414
# previous revision id or create a new revision? 20060606
415
if len(previous_entries) == 1:
416
# cannot be unchanged unless there is only one parent file rev.
417
parent_ie = previous_entries.values()[0]
418
if self._unchanged(parent_ie):
419
mutter("found unchanged entry")
420
self.revision = parent_ie.revision
422
return self._snapshot_into_revision(revision, previous_entries,
423
work_tree, commit_builder)
425
def _snapshot_into_revision(self, revision, previous_entries, work_tree,
427
"""Record this revision unconditionally into a store.
429
The entry's last-changed revision property (`revision`) is updated to
430
that of the new revision.
432
:param revision: id of the new revision that is being recorded.
434
:returns: String description of the commit (e.g. "merged", "modified"), etc.
436
mutter('new revision {%s} for {%s}', revision, self.file_id)
437
self.revision = revision
438
self._snapshot_text(previous_entries, work_tree, commit_builder)
440
def _snapshot_text(self, file_parents, work_tree, commit_builder):
441
"""Record the 'text' of this entry, whatever form that takes.
443
This default implementation simply adds an empty text.
445
raise NotImplementedError(self._snapshot_text)
447
def __eq__(self, other):
448
if not isinstance(other, InventoryEntry):
449
return NotImplemented
451
return ((self.file_id == other.file_id)
452
and (self.name == other.name)
453
and (other.symlink_target == self.symlink_target)
454
and (self.text_sha1 == other.text_sha1)
455
and (self.text_size == other.text_size)
456
and (self.text_id == other.text_id)
457
and (self.parent_id == other.parent_id)
458
and (self.kind == other.kind)
459
and (self.revision == other.revision)
460
and (self.executable == other.executable)
463
def __ne__(self, other):
464
return not (self == other)
467
raise ValueError('not hashable')
469
def _unchanged(self, previous_ie):
470
"""Has this entry changed relative to previous_ie.
472
This method should be overridden in child classes.
475
# different inv parent
476
if previous_ie.parent_id != self.parent_id:
479
elif previous_ie.name != self.name:
483
def _read_tree_state(self, path, work_tree):
484
"""Populate fields in the inventory entry from the given tree.
486
Note that this should be modified to be a noop on virtual trees
487
as all entries created there are prepopulated.
489
# TODO: Rather than running this manually, we should check the
490
# working sha1 and other expensive properties when they're
491
# first requested, or preload them if they're already known
492
pass # nothing to do by default
494
def _forget_tree_state(self):
498
class RootEntry(InventoryEntry):
500
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
501
'text_id', 'parent_id', 'children', 'executable',
502
'revision', 'symlink_target']
504
def _check(self, checker, rev_id, tree):
505
"""See InventoryEntry._check"""
507
def __init__(self, file_id):
508
self.file_id = file_id
510
self.kind = 'root_directory'
511
self.parent_id = None
515
def __eq__(self, other):
516
if not isinstance(other, RootEntry):
517
return NotImplemented
519
return (self.file_id == other.file_id) \
520
and (self.children == other.children)
523
class InventoryDirectory(InventoryEntry):
524
"""A directory in an inventory."""
526
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
527
'text_id', 'parent_id', 'children', 'executable',
528
'revision', 'symlink_target']
530
def _check(self, checker, rev_id, tree):
531
"""See InventoryEntry._check"""
532
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
533
raise BzrCheckError('directory {%s} has text in revision {%s}'
534
% (self.file_id, rev_id))
537
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
538
other.revision = self.revision
539
# note that children are *not* copied; they're pulled across when
543
def __init__(self, file_id, name, parent_id):
544
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
546
self.kind = 'directory'
548
def kind_character(self):
549
"""See InventoryEntry.kind_character."""
552
def _put_in_tar(self, item, tree):
553
"""See InventoryEntry._put_in_tar."""
554
item.type = tarfile.DIRTYPE
561
def _put_on_disk(self, fullpath, tree):
562
"""See InventoryEntry._put_on_disk."""
565
def _snapshot_text(self, file_parents, work_tree, commit_builder):
566
"""See InventoryEntry._snapshot_text."""
567
commit_builder.modified_directory(self.file_id, file_parents)
570
class InventoryFile(InventoryEntry):
571
"""A file in an inventory."""
573
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
574
'text_id', 'parent_id', 'children', 'executable',
575
'revision', 'symlink_target']
577
def _check(self, checker, tree_revision_id, tree):
578
"""See InventoryEntry._check"""
579
t = (self.file_id, self.revision)
580
if t in checker.checked_texts:
581
prev_sha = checker.checked_texts[t]
582
if prev_sha != self.text_sha1:
583
raise BzrCheckError('mismatched sha1 on {%s} in {%s}' %
584
(self.file_id, tree_revision_id))
586
checker.repeated_text_cnt += 1
589
if self.file_id not in checker.checked_weaves:
590
mutter('check weave {%s}', self.file_id)
591
w = tree.get_weave(self.file_id)
592
# Not passing a progress bar, because it creates a new
593
# progress, which overwrites the current progress,
594
# and doesn't look nice
596
checker.checked_weaves[self.file_id] = True
598
w = tree.get_weave(self.file_id)
600
mutter('check version {%s} of {%s}', tree_revision_id, self.file_id)
601
checker.checked_text_cnt += 1
602
# We can't check the length, because Weave doesn't store that
603
# information, and the whole point of looking at the weave's
604
# sha1sum is that we don't have to extract the text.
605
if self.text_sha1 != w.get_sha1(self.revision):
606
raise BzrCheckError('text {%s} version {%s} wrong sha1'
607
% (self.file_id, self.revision))
608
checker.checked_texts[t] = self.text_sha1
611
other = InventoryFile(self.file_id, self.name, self.parent_id)
612
other.executable = self.executable
613
other.text_id = self.text_id
614
other.text_sha1 = self.text_sha1
615
other.text_size = self.text_size
616
other.revision = self.revision
619
def detect_changes(self, old_entry):
620
"""See InventoryEntry.detect_changes."""
621
assert self.text_sha1 is not None
622
assert old_entry.text_sha1 is not None
623
text_modified = (self.text_sha1 != old_entry.text_sha1)
624
meta_modified = (self.executable != old_entry.executable)
625
return text_modified, meta_modified
627
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
628
output_to, reverse=False):
629
"""See InventoryEntry._diff."""
631
from_text = tree.get_file(self.file_id).readlines()
633
to_text = to_tree.get_file(to_entry.file_id).readlines()
637
text_diff(from_label, from_text,
638
to_label, to_text, output_to)
640
text_diff(to_label, to_text,
641
from_label, from_text, output_to)
644
label_pair = (to_label, from_label)
646
label_pair = (from_label, to_label)
647
print >> output_to, "Binary files %s and %s differ" % label_pair
650
"""See InventoryEntry.has_text."""
653
def __init__(self, file_id, name, parent_id):
654
super(InventoryFile, self).__init__(file_id, name, parent_id)
657
def kind_character(self):
658
"""See InventoryEntry.kind_character."""
661
def _put_in_tar(self, item, tree):
662
"""See InventoryEntry._put_in_tar."""
663
item.type = tarfile.REGTYPE
664
fileobj = tree.get_file(self.file_id)
665
item.size = self.text_size
666
if tree.is_executable(self.file_id):
672
def _put_on_disk(self, fullpath, tree):
673
"""See InventoryEntry._put_on_disk."""
674
pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
675
if tree.is_executable(self.file_id):
676
os.chmod(fullpath, 0755)
678
def _read_tree_state(self, path, work_tree):
679
"""See InventoryEntry._read_tree_state."""
680
self.text_sha1 = work_tree.get_file_sha1(self.file_id, path=path)
681
# FIXME: 20050930 probe for the text size when getting sha1
682
# in _read_tree_state
683
self.executable = work_tree.is_executable(self.file_id, path=path)
686
return ("%s(%r, %r, parent_id=%r, sha1=%r, len=%s)"
687
% (self.__class__.__name__,
694
def _forget_tree_state(self):
695
self.text_sha1 = None
697
def _snapshot_text(self, file_parents, work_tree, commit_builder):
698
"""See InventoryEntry._snapshot_text."""
699
def get_content_byte_lines():
700
return work_tree.get_file(self.file_id).readlines()
701
self.text_sha1, self.text_size = commit_builder.modified_file_text(
702
self.file_id, file_parents, get_content_byte_lines, self.text_sha1, self.text_size)
704
def _unchanged(self, previous_ie):
705
"""See InventoryEntry._unchanged."""
706
compatible = super(InventoryFile, self)._unchanged(previous_ie)
707
if self.text_sha1 != previous_ie.text_sha1:
710
# FIXME: 20050930 probe for the text size when getting sha1
711
# in _read_tree_state
712
self.text_size = previous_ie.text_size
713
if self.executable != previous_ie.executable:
718
class InventoryLink(InventoryEntry):
719
"""A file in an inventory."""
721
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
722
'text_id', 'parent_id', 'children', 'executable',
723
'revision', 'symlink_target']
725
def _check(self, checker, rev_id, tree):
726
"""See InventoryEntry._check"""
727
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
728
raise BzrCheckError('symlink {%s} has text in revision {%s}'
729
% (self.file_id, rev_id))
730
if self.symlink_target is None:
731
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
732
% (self.file_id, rev_id))
735
other = InventoryLink(self.file_id, self.name, self.parent_id)
736
other.symlink_target = self.symlink_target
737
other.revision = self.revision
740
def detect_changes(self, old_entry):
741
"""See InventoryEntry.detect_changes."""
742
# FIXME: which _modified field should we use ? RBC 20051003
743
text_modified = (self.symlink_target != old_entry.symlink_target)
745
mutter(" symlink target changed")
746
meta_modified = False
747
return text_modified, meta_modified
749
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
750
output_to, reverse=False):
751
"""See InventoryEntry._diff."""
752
from_text = self.symlink_target
753
if to_entry is not None:
754
to_text = to_entry.symlink_target
759
print >>output_to, '=== target changed %r => %r' % (from_text, to_text)
762
print >>output_to, '=== target was %r' % self.symlink_target
764
print >>output_to, '=== target is %r' % self.symlink_target
766
def __init__(self, file_id, name, parent_id):
767
super(InventoryLink, self).__init__(file_id, name, parent_id)
768
self.kind = 'symlink'
770
def kind_character(self):
771
"""See InventoryEntry.kind_character."""
774
def _put_in_tar(self, item, tree):
775
"""See InventoryEntry._put_in_tar."""
776
item.type = tarfile.SYMTYPE
780
item.linkname = self.symlink_target
783
def _put_on_disk(self, fullpath, tree):
784
"""See InventoryEntry._put_on_disk."""
786
os.symlink(self.symlink_target, fullpath)
788
raise BzrError("Failed to create symlink %r -> %r, error: %s" % (fullpath, self.symlink_target, e))
790
def _read_tree_state(self, path, work_tree):
791
"""See InventoryEntry._read_tree_state."""
792
self.symlink_target = work_tree.get_symlink_target(self.file_id)
794
def _forget_tree_state(self):
795
self.symlink_target = None
797
def _unchanged(self, previous_ie):
798
"""See InventoryEntry._unchanged."""
799
compatible = super(InventoryLink, self)._unchanged(previous_ie)
800
if self.symlink_target != previous_ie.symlink_target:
804
def _snapshot_text(self, file_parents, work_tree, commit_builder):
805
"""See InventoryEntry._snapshot_text."""
806
commit_builder.modified_link(
807
self.file_id, file_parents, self.symlink_target)
810
class Inventory(object):
811
"""Inventory of versioned files in a tree.
813
This describes which file_id is present at each point in the tree,
814
and possibly the SHA-1 or other information about the file.
815
Entries can be looked up either by path or by file_id.
817
The inventory represents a typical unix file tree, with
818
directories containing files and subdirectories. We never store
819
the full path to a file, because renaming a directory implicitly
820
moves all of its contents. This class internally maintains a
821
lookup tree that allows the children under a directory to be
824
InventoryEntry objects must not be modified after they are
825
inserted, other than through the Inventory API.
827
>>> inv = Inventory()
828
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
829
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
830
>>> inv['123-123'].name
833
May be treated as an iterator or set to look up file ids:
835
>>> bool(inv.path2id('hello.c'))
840
May also look up by name:
842
>>> [x[0] for x in inv.iter_entries()]
844
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
845
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
846
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT-12345678-12345678', sha1=None, len=None)
848
def __init__(self, root_id=ROOT_ID, revision_id=None):
849
"""Create or read an inventory.
851
If a working directory is specified, the inventory is read
852
from there. If the file is specified, read from that. If not,
853
the inventory is created empty.
855
The inventory is created with a default root directory, with
858
# We are letting Branch.create() create a unique inventory
859
# root id. Rather than generating a random one here.
861
# root_id = bzrlib.branch.gen_file_id('TREE_ROOT')
862
self.root = RootEntry(root_id)
863
# FIXME: this isn't ever used, changing it to self.revision may break
864
# things. TODO make everything use self.revision_id
865
self.revision_id = revision_id
866
self._byid = {self.root.file_id: self.root}
869
# TODO: jam 20051218 Should copy also copy the revision_id?
870
other = Inventory(self.root.file_id)
871
# copy recursively so we know directories will be added before
872
# their children. There are more efficient ways than this...
873
for path, entry in self.iter_entries():
874
if entry == self.root:
876
other.add(entry.copy())
880
return iter(self._byid)
883
"""Returns number of entries."""
884
return len(self._byid)
886
def iter_entries(self, from_dir=None):
887
"""Return (path, entry) pairs, in order by name."""
891
elif isinstance(from_dir, basestring):
892
from_dir = self._byid[from_dir]
894
# unrolling the recursive called changed the time from
895
# 440ms/663ms (inline/total) to 116ms/116ms
896
children = from_dir.children.items()
898
children = collections.deque(children)
899
stack = [(u'', children)]
901
from_dir_relpath, children = stack[-1]
904
name, ie = children.popleft()
906
# we know that from_dir_relpath never ends in a slash
907
# and 'f' doesn't begin with one, we can do a string op, rather
908
# than the checks of pathjoin(), though this means that all paths
910
path = from_dir_relpath + '/' + name
914
if ie.kind != 'directory':
917
# But do this child first
918
new_children = ie.children.items()
920
new_children = collections.deque(new_children)
921
stack.append((path, new_children))
922
# Break out of inner loop, so that we start outer loop with child
925
# if we finished all children, pop it off the stack
928
def iter_entries_by_dir(self, from_dir=None):
929
"""Iterate over the entries in a directory first order.
931
This returns all entries for a directory before returning
932
the entries for children of a directory. This is not
933
lexicographically sorted order, and is a hybrid between
934
depth-first and breadth-first.
936
:return: This yields (path, entry) pairs
938
# TODO? Perhaps this should return the from_dir so that the root is
939
# yielded? or maybe an option?
943
elif isinstance(from_dir, basestring):
944
from_dir = self._byid[from_dir]
946
stack = [(u'', from_dir)]
948
cur_relpath, cur_dir = stack.pop()
951
for child_name, child_ie in sorted(cur_dir.children.iteritems()):
953
child_relpath = cur_relpath + child_name
955
yield child_relpath, child_ie
957
if child_ie.kind == 'directory':
958
child_dirs.append((child_relpath+'/', child_ie))
959
stack.extend(reversed(child_dirs))
962
"""Return list of (path, ie) for all entries except the root.
964
This may be faster than iter_entries.
967
def descend(dir_ie, dir_path):
968
kids = dir_ie.children.items()
970
for name, ie in kids:
971
child_path = pathjoin(dir_path, name)
972
accum.append((child_path, ie))
973
if ie.kind == 'directory':
974
descend(ie, child_path)
976
descend(self.root, u'')
979
def directories(self):
980
"""Return (path, entry) pairs for all directories, including the root.
983
def descend(parent_ie, parent_path):
984
accum.append((parent_path, parent_ie))
986
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
989
for name, child_ie in kids:
990
child_path = pathjoin(parent_path, name)
991
descend(child_ie, child_path)
992
descend(self.root, u'')
995
def __contains__(self, file_id):
996
"""True if this entry contains a file with given id.
998
>>> inv = Inventory()
999
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1000
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
1006
return file_id in self._byid
1008
def __getitem__(self, file_id):
1009
"""Return the entry for given file_id.
1011
>>> inv = Inventory()
1012
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
1013
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
1014
>>> inv['123123'].name
1018
return self._byid[file_id]
1021
raise BzrError("can't look up file_id None")
1023
raise BzrError("file_id {%s} not in inventory" % file_id)
1025
def get_file_kind(self, file_id):
1026
return self._byid[file_id].kind
1028
def get_child(self, parent_id, filename):
1029
return self[parent_id].children.get(filename)
1031
def add(self, entry):
1032
"""Add entry to inventory.
1034
To add a file to a branch ready to be committed, use Branch.add,
1037
Returns the new entry object.
1039
if entry.file_id in self._byid:
1040
raise BzrError("inventory already contains entry with id {%s}" % entry.file_id)
1042
if entry.parent_id == ROOT_ID or entry.parent_id is None:
1043
entry.parent_id = self.root.file_id
1046
parent = self._byid[entry.parent_id]
1048
raise BzrError("parent_id {%s} not in inventory" % entry.parent_id)
1050
if entry.name in parent.children:
1051
raise BzrError("%s is already versioned" %
1052
pathjoin(self.id2path(parent.file_id), entry.name))
1054
self._byid[entry.file_id] = entry
1055
parent.children[entry.name] = entry
1058
def add_path(self, relpath, kind, file_id=None, parent_id=None):
1059
"""Add entry from a path.
1061
The immediate parent must already be versioned.
1063
Returns the new entry object."""
1065
parts = osutils.splitpath(relpath)
1069
file_id = bzrlib.workingtree.gen_root_id()
1070
self.root = RootEntry(file_id)
1071
self._byid = {self.root.file_id: self.root}
1074
parent_path = parts[:-1]
1075
parent_id = self.path2id(parent_path)
1076
if parent_id is None:
1077
raise NotVersionedError(path=parent_path)
1078
ie = make_entry(kind, parts[-1], parent_id, file_id)
1081
def __delitem__(self, file_id):
1082
"""Remove entry by id.
1084
>>> inv = Inventory()
1085
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1086
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
1095
assert ie.parent_id is None or \
1096
self[ie.parent_id].children[ie.name] == ie
1098
del self._byid[file_id]
1099
if ie.parent_id is not None:
1100
del self[ie.parent_id].children[ie.name]
1102
def __eq__(self, other):
1103
"""Compare two sets by comparing their contents.
1105
>>> i1 = Inventory()
1106
>>> i2 = Inventory()
1109
>>> i1.add(InventoryFile('123', 'foo', ROOT_ID))
1110
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
1113
>>> i2.add(InventoryFile('123', 'foo', ROOT_ID))
1114
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
1118
if not isinstance(other, Inventory):
1119
return NotImplemented
1121
return self._byid == other._byid
1123
def __ne__(self, other):
1124
return not self.__eq__(other)
1127
raise ValueError('not hashable')
1129
def _iter_file_id_parents(self, file_id):
1130
"""Yield the parents of file_id up to the root."""
1131
while file_id is not None:
1133
ie = self._byid[file_id]
1135
raise BzrError("file_id {%s} not found in inventory" % file_id)
1137
file_id = ie.parent_id
1139
def get_idpath(self, file_id):
1140
"""Return a list of file_ids for the path to an entry.
1142
The list contains one element for each directory followed by
1143
the id of the file itself. So the length of the returned list
1144
is equal to the depth of the file in the tree, counting the
1145
root directory as depth 1.
1148
for parent in self._iter_file_id_parents(file_id):
1149
p.insert(0, parent.file_id)
1152
def id2path(self, file_id):
1153
"""Return as a string the path to file_id.
1156
>>> e = i.add(InventoryDirectory('src-id', 'src', ROOT_ID))
1157
>>> e = i.add(InventoryFile('foo-id', 'foo.c', parent_id='src-id'))
1158
>>> print i.id2path('foo-id')
1161
# get all names, skipping root
1162
return '/'.join(reversed(
1163
[parent.name for parent in
1164
self._iter_file_id_parents(file_id)][:-1]))
1166
def path2id(self, name):
1167
"""Walk down through directories to return entry of last component.
1169
names may be either a list of path components, or a single
1170
string, in which case it is automatically split.
1172
This returns the entry of the last component in the path,
1173
which may be either a file or a directory.
1175
Returns None IFF the path is not found.
1177
if isinstance(name, types.StringTypes):
1178
name = splitpath(name)
1180
# mutter("lookup path %r" % name)
1185
cie = parent.children[f]
1186
assert cie.name == f
1187
assert cie.parent_id == parent.file_id
1190
# or raise an error?
1193
return parent.file_id
1195
def has_filename(self, names):
1196
return bool(self.path2id(names))
1198
def has_id(self, file_id):
1199
return self._byid.has_key(file_id)
1201
def rename(self, file_id, new_parent_id, new_name):
1202
"""Move a file within the inventory.
1204
This can change either the name, or the parent, or both.
1206
This does not move the working file."""
1207
if not is_valid_name(new_name):
1208
raise BzrError("not an acceptable filename: %r" % new_name)
1210
new_parent = self._byid[new_parent_id]
1211
if new_name in new_parent.children:
1212
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
1214
new_parent_idpath = self.get_idpath(new_parent_id)
1215
if file_id in new_parent_idpath:
1216
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
1217
% (self.id2path(file_id), self.id2path(new_parent_id)))
1219
file_ie = self._byid[file_id]
1220
old_parent = self._byid[file_ie.parent_id]
1222
# TODO: Don't leave things messed up if this fails
1224
del old_parent.children[file_ie.name]
1225
new_parent.children[new_name] = file_ie
1227
file_ie.name = new_name
1228
file_ie.parent_id = new_parent_id
1231
def make_entry(kind, name, parent_id, file_id=None):
1232
"""Create an inventory entry.
1234
:param kind: the type of inventory entry to create.
1235
:param name: the basename of the entry.
1236
:param parent_id: the parent_id of the entry.
1237
:param file_id: the file_id to use. if None, one will be created.
1240
file_id = bzrlib.workingtree.gen_file_id(name)
1242
norm_name, can_access = osutils.normalized_filename(name)
1243
if norm_name != name:
1247
# TODO: jam 20060701 This would probably be more useful
1248
# if the error was raised with the full path
1249
raise errors.InvalidNormalization(name)
1251
if kind == 'directory':
1252
return InventoryDirectory(file_id, name, parent_id)
1253
elif kind == 'file':
1254
return InventoryFile(file_id, name, parent_id)
1255
elif kind == 'symlink':
1256
return InventoryLink(file_id, name, parent_id)
1258
raise BzrError("unknown kind %r" % kind)
1263
def is_valid_name(name):
1265
if _NAME_RE is None:
1266
_NAME_RE = re.compile(r'^[^/\\]+$')
1268
return bool(_NAME_RE.match(name))
added
removed
bzrlib/inventory.py
