1
# Copyright (C) 2005, 2006, 2007 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.
34
from bzrlib.lazy_import import lazy_import
35
lazy_import(globals(), """
49
from bzrlib.errors import (
53
from bzrlib.trace import mutter
56
class InventoryEntry(object):
57
"""Description of a versioned file.
59
An InventoryEntry has the following fields, which are also
60
present in the XML inventory-entry element:
65
(within the parent directory)
68
file_id of the parent directory, or ROOT_ID
71
the revision_id in which this variation of this file was
75
Indicates that this file should be executable on systems
79
sha-1 of the text of the file
82
size in bytes of the text of the file
84
(reading a version 4 tree created a text_id field.)
89
>>> i.add(InventoryDirectory('123', 'src', ROOT_ID))
90
InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None)
91
>>> i.add(InventoryFile('2323', 'hello.c', parent_id='123'))
92
InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None)
93
>>> shouldbe = {0: '', 1: 'src', 2: 'src/hello.c'}
94
>>> for ix, j in enumerate(i.iter_entries()):
95
... print (j[0] == shouldbe[ix], j[1])
97
(True, InventoryDirectory('TREE_ROOT', u'', parent_id=None, revision=None))
98
(True, InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None))
99
(True, InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None))
100
>>> i.add(InventoryFile('2324', 'bye.c', '123'))
101
InventoryFile('2324', 'bye.c', parent_id='123', sha1=None, len=None)
102
>>> i.add(InventoryDirectory('2325', 'wibble', '123'))
103
InventoryDirectory('2325', 'wibble', parent_id='123', revision=None)
104
>>> i.path2id('src/wibble')
108
>>> i.add(InventoryFile('2326', 'wibble.c', '2325'))
109
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
111
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
112
>>> for path, entry in i.iter_entries():
114
... assert i.path2id(path)
122
>>> i.id2path('2326')
123
'src/wibble/wibble.c'
126
# Constants returned by describe_change()
128
# TODO: These should probably move to some kind of FileChangeDescription
129
# class; that's like what's inside a TreeDelta but we want to be able to
130
# generate them just for one file at a time.
132
MODIFIED_AND_RENAMED = 'modified and renamed'
136
def detect_changes(self, old_entry):
137
"""Return a (text_modified, meta_modified) from this to old_entry.
139
_read_tree_state must have been called on self and old_entry prior to
140
calling detect_changes.
144
def diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
145
output_to, reverse=False):
146
"""Perform a diff from this to to_entry.
148
text_diff will be used for textual difference calculation.
149
This is a template method, override _diff in child classes.
151
self._read_tree_state(tree.id2path(self.file_id), tree)
153
# cannot diff from one kind to another - you must do a removal
154
# and an addif they do not match.
155
assert self.kind == to_entry.kind
156
to_entry._read_tree_state(to_tree.id2path(to_entry.file_id),
158
self._diff(text_diff, from_label, tree, to_label, to_entry, to_tree,
161
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
162
output_to, reverse=False):
163
"""Perform a diff between two entries of the same kind."""
165
def find_previous_heads(self, previous_inventories,
166
versioned_file_store,
169
"""Return the revisions and entries that directly precede this.
171
Returned as a map from revision to inventory entry.
173
This is a map containing the file revisions in all parents
174
for which the file exists, and its revision is not a parent of
175
any other. If the file is new, the set will be empty.
177
:param versioned_file_store: A store where ancestry data on this
178
file id can be queried.
179
:param transaction: The transaction that queries to the versioned
180
file store should be completed under.
181
:param entry_vf: The entry versioned file, if its already available.
183
def get_ancestors(weave, entry):
184
return set(weave.get_ancestry(entry.revision))
185
# revision:ie mapping for each ie found in previous_inventories.
187
# revision:ie mapping with one revision for each head.
189
# revision: ancestor list for each head
191
# identify candidate head revision ids.
192
for inv in previous_inventories:
193
if self.file_id in inv:
194
ie = inv[self.file_id]
195
assert ie.file_id == self.file_id
196
if ie.revision in candidates:
197
# same revision value in two different inventories:
198
# correct possible inconsistencies:
199
# * there was a bug in revision updates with 'x' bit
202
if candidates[ie.revision].executable != ie.executable:
203
candidates[ie.revision].executable = False
204
ie.executable = False
205
except AttributeError:
207
# must now be the same.
208
assert candidates[ie.revision] == ie
210
# add this revision as a candidate.
211
candidates[ie.revision] = ie
213
# common case optimisation
214
if len(candidates) == 1:
215
# if there is only one candidate revision found
216
# then we can opening the versioned file to access ancestry:
217
# there cannot be any ancestors to eliminate when there is
218
# only one revision available.
219
heads[ie.revision] = ie
222
# eliminate ancestors amongst the available candidates:
223
# heads are those that are not an ancestor of any other candidate
224
# - this provides convergence at a per-file level.
225
for ie in candidates.values():
226
# may be an ancestor of a known head:
227
already_present = 0 != len(
228
[head for head in heads
229
if ie.revision in head_ancestors[head]])
231
# an ancestor of an analyzed candidate.
233
# not an ancestor of a known head:
234
# load the versioned file for this file id if needed
236
entry_vf = versioned_file_store.get_weave_or_empty(
237
self.file_id, transaction)
238
ancestors = get_ancestors(entry_vf, ie)
239
# may knock something else out:
240
check_heads = list(heads.keys())
241
for head in check_heads:
242
if head in ancestors:
243
# this previously discovered 'head' is not
244
# really a head - its an ancestor of the newly
247
head_ancestors[ie.revision] = ancestors
248
heads[ie.revision] = ie
251
def get_tar_item(self, root, dp, now, tree):
252
"""Get a tarfile item and a file stream for its content."""
253
item = tarfile.TarInfo(osutils.pathjoin(root, dp).encode('utf8'))
254
# TODO: would be cool to actually set it to the timestamp of the
255
# revision it was last changed
257
fileobj = self._put_in_tar(item, tree)
261
"""Return true if the object this entry represents has textual data.
263
Note that textual data includes binary content.
265
Also note that all entries get weave files created for them.
266
This attribute is primarily used when upgrading from old trees that
267
did not have the weave index for all inventory entries.
271
def __init__(self, file_id, name, parent_id, text_id=None):
272
"""Create an InventoryEntry
274
The filename must be a single component, relative to the
275
parent directory; it cannot be a whole path or relative name.
277
>>> e = InventoryFile('123', 'hello.c', ROOT_ID)
282
>>> e = InventoryFile('123', 'src/hello.c', ROOT_ID)
283
Traceback (most recent call last):
284
InvalidEntryName: Invalid entry name: src/hello.c
286
assert isinstance(name, basestring), name
287
if '/' in name or '\\' in name:
288
raise errors.InvalidEntryName(name=name)
289
self.executable = False
291
self.text_sha1 = None
292
self.text_size = None
293
self.file_id = file_id
294
assert isinstance(file_id, (str, None.__class__)), \
295
'bad type %r for %r' % (type(file_id), file_id)
297
self.text_id = text_id
298
self.parent_id = parent_id
299
self.symlink_target = None
300
self.reference_revision = None
302
def kind_character(self):
303
"""Return a short kind indicator useful for appending to names."""
304
raise BzrError('unknown kind %r' % self.kind)
306
known_kinds = ('file', 'directory', 'symlink')
308
def _put_in_tar(self, item, tree):
309
"""populate item for stashing in a tar, and return the content stream.
311
If no content is available, return None.
313
raise BzrError("don't know how to export {%s} of kind %r" %
314
(self.file_id, self.kind))
316
def put_on_disk(self, dest, dp, tree):
317
"""Create a representation of self on disk in the prefix dest.
319
This is a template method - implement _put_on_disk in subclasses.
321
fullpath = osutils.pathjoin(dest, dp)
322
self._put_on_disk(fullpath, tree)
323
# mutter(" export {%s} kind %s to %s", self.file_id,
324
# self.kind, fullpath)
326
def _put_on_disk(self, fullpath, tree):
327
"""Put this entry onto disk at fullpath, from tree tree."""
328
raise BzrError("don't know how to export {%s} of kind %r" % (self.file_id, self.kind))
330
def sorted_children(self):
331
return sorted(self.children.items())
334
def versionable_kind(kind):
335
return (kind in ('file', 'directory', 'symlink', 'tree-reference'))
337
def check(self, checker, rev_id, inv, tree):
338
"""Check this inventory entry is intact.
340
This is a template method, override _check for kind specific
343
:param checker: Check object providing context for the checks;
344
can be used to find out what parts of the repository have already
346
:param rev_id: Revision id from which this InventoryEntry was loaded.
347
Not necessarily the last-changed revision for this file.
348
:param inv: Inventory from which the entry was loaded.
349
:param tree: RevisionTree for this entry.
351
if self.parent_id is not None:
352
if not inv.has_id(self.parent_id):
353
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
354
% (self.parent_id, rev_id))
355
self._check(checker, rev_id, tree)
357
def _check(self, checker, rev_id, tree):
358
"""Check this inventory entry for kind specific errors."""
359
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
363
"""Clone this inventory entry."""
364
raise NotImplementedError
367
def describe_change(old_entry, new_entry):
368
"""Describe the change between old_entry and this.
370
This smells of being an InterInventoryEntry situation, but as its
371
the first one, we're making it a static method for now.
373
An entry with a different parent, or different name is considered
374
to be renamed. Reparenting is an internal detail.
375
Note that renaming the parent does not trigger a rename for the
378
# TODO: Perhaps return an object rather than just a string
379
if old_entry is new_entry:
380
# also the case of both being None
382
elif old_entry is None:
384
elif new_entry is None:
386
if old_entry.kind != new_entry.kind:
388
text_modified, meta_modified = new_entry.detect_changes(old_entry)
389
if text_modified or meta_modified:
393
# TODO 20060511 (mbp, rbc) factor out 'detect_rename' here.
394
if old_entry.parent_id != new_entry.parent_id:
396
elif old_entry.name != new_entry.name:
400
if renamed and not modified:
401
return InventoryEntry.RENAMED
402
if modified and not renamed:
404
if modified and renamed:
405
return InventoryEntry.MODIFIED_AND_RENAMED
409
return ("%s(%r, %r, parent_id=%r, revision=%r)"
410
% (self.__class__.__name__,
416
def snapshot(self, revision, path, previous_entries,
417
work_tree, commit_builder):
418
"""Make a snapshot of this entry which may or may not have changed.
420
This means that all its fields are populated, that it has its
421
text stored in the text store or weave.
423
# mutter('new parents of %s are %r', path, previous_entries)
424
self._read_tree_state(path, work_tree)
425
# TODO: Where should we determine whether to reuse a
426
# previous revision id or create a new revision? 20060606
427
if len(previous_entries) == 1:
428
# cannot be unchanged unless there is only one parent file rev.
429
parent_ie = previous_entries.values()[0]
430
if self._unchanged(parent_ie):
431
# mutter("found unchanged entry")
432
self.revision = parent_ie.revision
434
return self._snapshot_into_revision(revision, previous_entries,
435
work_tree, commit_builder)
437
def _snapshot_into_revision(self, revision, previous_entries, work_tree,
439
"""Record this revision unconditionally into a store.
441
The entry's last-changed revision property (`revision`) is updated to
442
that of the new revision.
444
:param revision: id of the new revision that is being recorded.
446
:returns: String description of the commit (e.g. "merged", "modified"), etc.
448
# mutter('new revision {%s} for {%s}', revision, self.file_id)
449
self.revision = revision
450
self._snapshot_text(previous_entries, work_tree, commit_builder)
452
def _snapshot_text(self, file_parents, work_tree, commit_builder):
453
"""Record the 'text' of this entry, whatever form that takes.
455
This default implementation simply adds an empty text.
457
raise NotImplementedError(self._snapshot_text)
459
def __eq__(self, other):
460
if not isinstance(other, InventoryEntry):
461
return NotImplemented
463
return ((self.file_id == other.file_id)
464
and (self.name == other.name)
465
and (other.symlink_target == self.symlink_target)
466
and (self.text_sha1 == other.text_sha1)
467
and (self.text_size == other.text_size)
468
and (self.text_id == other.text_id)
469
and (self.parent_id == other.parent_id)
470
and (self.kind == other.kind)
471
and (self.revision == other.revision)
472
and (self.executable == other.executable)
473
and (self.reference_revision == other.reference_revision)
476
def __ne__(self, other):
477
return not (self == other)
480
raise ValueError('not hashable')
482
def _unchanged(self, previous_ie):
483
"""Has this entry changed relative to previous_ie.
485
This method should be overridden in child classes.
488
# different inv parent
489
if previous_ie.parent_id != self.parent_id:
492
elif previous_ie.name != self.name:
496
def _read_tree_state(self, path, work_tree):
497
"""Populate fields in the inventory entry from the given tree.
499
Note that this should be modified to be a noop on virtual trees
500
as all entries created there are prepopulated.
502
# TODO: Rather than running this manually, we should check the
503
# working sha1 and other expensive properties when they're
504
# first requested, or preload them if they're already known
505
pass # nothing to do by default
507
def _forget_tree_state(self):
511
class RootEntry(InventoryEntry):
513
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
514
'text_id', 'parent_id', 'children', 'executable',
515
'revision', 'symlink_target', 'reference_revision']
517
def _check(self, checker, rev_id, tree):
518
"""See InventoryEntry._check"""
520
def __init__(self, file_id):
521
self.file_id = file_id
523
self.kind = 'directory'
524
self.parent_id = None
527
symbol_versioning.warn('RootEntry is deprecated as of bzr 0.10.'
528
' Please use InventoryDirectory instead.',
529
DeprecationWarning, stacklevel=2)
531
def __eq__(self, other):
532
if not isinstance(other, RootEntry):
533
return NotImplemented
535
return (self.file_id == other.file_id) \
536
and (self.children == other.children)
539
class InventoryDirectory(InventoryEntry):
540
"""A directory in an inventory."""
542
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
543
'text_id', 'parent_id', 'children', 'executable',
544
'revision', 'symlink_target', 'reference_revision']
546
def _check(self, checker, rev_id, tree):
547
"""See InventoryEntry._check"""
548
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
549
raise BzrCheckError('directory {%s} has text in revision {%s}'
550
% (self.file_id, rev_id))
553
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
554
other.revision = self.revision
555
# note that children are *not* copied; they're pulled across when
559
def __init__(self, file_id, name, parent_id):
560
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
562
self.kind = 'directory'
564
def kind_character(self):
565
"""See InventoryEntry.kind_character."""
568
def _put_in_tar(self, item, tree):
569
"""See InventoryEntry._put_in_tar."""
570
item.type = tarfile.DIRTYPE
577
def _put_on_disk(self, fullpath, tree):
578
"""See InventoryEntry._put_on_disk."""
581
def _snapshot_text(self, file_parents, work_tree, commit_builder):
582
"""See InventoryEntry._snapshot_text."""
583
commit_builder.modified_directory(self.file_id, file_parents)
586
class InventoryFile(InventoryEntry):
587
"""A file in an inventory."""
589
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
590
'text_id', 'parent_id', 'children', 'executable',
591
'revision', 'symlink_target', 'reference_revision']
593
def _check(self, checker, tree_revision_id, tree):
594
"""See InventoryEntry._check"""
595
t = (self.file_id, self.revision)
596
if t in checker.checked_texts:
597
prev_sha = checker.checked_texts[t]
598
if prev_sha != self.text_sha1:
599
raise BzrCheckError('mismatched sha1 on {%s} in {%s}' %
600
(self.file_id, tree_revision_id))
602
checker.repeated_text_cnt += 1
605
if self.file_id not in checker.checked_weaves:
606
mutter('check weave {%s}', self.file_id)
607
w = tree.get_weave(self.file_id)
608
# Not passing a progress bar, because it creates a new
609
# progress, which overwrites the current progress,
610
# and doesn't look nice
612
checker.checked_weaves[self.file_id] = True
614
w = tree.get_weave(self.file_id)
616
mutter('check version {%s} of {%s}', tree_revision_id, self.file_id)
617
checker.checked_text_cnt += 1
618
# We can't check the length, because Weave doesn't store that
619
# information, and the whole point of looking at the weave's
620
# sha1sum is that we don't have to extract the text.
621
if self.text_sha1 != w.get_sha1(self.revision):
622
raise BzrCheckError('text {%s} version {%s} wrong sha1'
623
% (self.file_id, self.revision))
624
checker.checked_texts[t] = self.text_sha1
627
other = InventoryFile(self.file_id, self.name, self.parent_id)
628
other.executable = self.executable
629
other.text_id = self.text_id
630
other.text_sha1 = self.text_sha1
631
other.text_size = self.text_size
632
other.revision = self.revision
635
def detect_changes(self, old_entry):
636
"""See InventoryEntry.detect_changes."""
637
assert self.text_sha1 is not None
638
assert old_entry.text_sha1 is not None
639
text_modified = (self.text_sha1 != old_entry.text_sha1)
640
meta_modified = (self.executable != old_entry.executable)
641
return text_modified, meta_modified
643
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
644
output_to, reverse=False):
645
"""See InventoryEntry._diff."""
647
from_text = tree.get_file(self.file_id).readlines()
649
to_text = to_tree.get_file(to_entry.file_id).readlines()
653
text_diff(from_label, from_text,
654
to_label, to_text, output_to)
656
text_diff(to_label, to_text,
657
from_label, from_text, output_to)
658
except errors.BinaryFile:
660
label_pair = (to_label, from_label)
662
label_pair = (from_label, to_label)
663
print >> output_to, "Binary files %s and %s differ" % label_pair
666
"""See InventoryEntry.has_text."""
669
def __init__(self, file_id, name, parent_id):
670
super(InventoryFile, self).__init__(file_id, name, parent_id)
673
def kind_character(self):
674
"""See InventoryEntry.kind_character."""
677
def _put_in_tar(self, item, tree):
678
"""See InventoryEntry._put_in_tar."""
679
item.type = tarfile.REGTYPE
680
fileobj = tree.get_file(self.file_id)
681
item.size = self.text_size
682
if tree.is_executable(self.file_id):
688
def _put_on_disk(self, fullpath, tree):
689
"""See InventoryEntry._put_on_disk."""
690
osutils.pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
691
if tree.is_executable(self.file_id):
692
os.chmod(fullpath, 0755)
694
def _read_tree_state(self, path, work_tree):
695
"""See InventoryEntry._read_tree_state."""
696
self.text_sha1 = work_tree.get_file_sha1(self.file_id, path=path)
697
# FIXME: 20050930 probe for the text size when getting sha1
698
# in _read_tree_state
699
self.executable = work_tree.is_executable(self.file_id, path=path)
702
return ("%s(%r, %r, parent_id=%r, sha1=%r, len=%s)"
703
% (self.__class__.__name__,
710
def _forget_tree_state(self):
711
self.text_sha1 = None
713
def _snapshot_text(self, file_parents, work_tree, commit_builder):
714
"""See InventoryEntry._snapshot_text."""
715
def get_content_byte_lines():
716
return work_tree.get_file(self.file_id).readlines()
717
self.text_sha1, self.text_size = commit_builder.modified_file_text(
718
self.file_id, file_parents, get_content_byte_lines, self.text_sha1, self.text_size)
720
def _unchanged(self, previous_ie):
721
"""See InventoryEntry._unchanged."""
722
compatible = super(InventoryFile, self)._unchanged(previous_ie)
723
if self.text_sha1 != previous_ie.text_sha1:
726
# FIXME: 20050930 probe for the text size when getting sha1
727
# in _read_tree_state
728
self.text_size = previous_ie.text_size
729
if self.executable != previous_ie.executable:
734
class InventoryLink(InventoryEntry):
735
"""A file in an inventory."""
737
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
738
'text_id', 'parent_id', 'children', 'executable',
739
'revision', 'symlink_target', 'reference_revision']
741
def _check(self, checker, rev_id, tree):
742
"""See InventoryEntry._check"""
743
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
744
raise BzrCheckError('symlink {%s} has text in revision {%s}'
745
% (self.file_id, rev_id))
746
if self.symlink_target is None:
747
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
748
% (self.file_id, rev_id))
751
other = InventoryLink(self.file_id, self.name, self.parent_id)
752
other.symlink_target = self.symlink_target
753
other.revision = self.revision
756
def detect_changes(self, old_entry):
757
"""See InventoryEntry.detect_changes."""
758
# FIXME: which _modified field should we use ? RBC 20051003
759
text_modified = (self.symlink_target != old_entry.symlink_target)
761
mutter(" symlink target changed")
762
meta_modified = False
763
return text_modified, meta_modified
765
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
766
output_to, reverse=False):
767
"""See InventoryEntry._diff."""
768
from_text = self.symlink_target
769
if to_entry is not None:
770
to_text = to_entry.symlink_target
775
print >>output_to, '=== target changed %r => %r' % (from_text, to_text)
778
print >>output_to, '=== target was %r' % self.symlink_target
780
print >>output_to, '=== target is %r' % self.symlink_target
782
def __init__(self, file_id, name, parent_id):
783
super(InventoryLink, self).__init__(file_id, name, parent_id)
784
self.kind = 'symlink'
786
def kind_character(self):
787
"""See InventoryEntry.kind_character."""
790
def _put_in_tar(self, item, tree):
791
"""See InventoryEntry._put_in_tar."""
792
item.type = tarfile.SYMTYPE
796
item.linkname = self.symlink_target
799
def _put_on_disk(self, fullpath, tree):
800
"""See InventoryEntry._put_on_disk."""
802
os.symlink(self.symlink_target, fullpath)
804
raise BzrError("Failed to create symlink %r -> %r, error: %s" % (fullpath, self.symlink_target, e))
806
def _read_tree_state(self, path, work_tree):
807
"""See InventoryEntry._read_tree_state."""
808
self.symlink_target = work_tree.get_symlink_target(self.file_id)
810
def _forget_tree_state(self):
811
self.symlink_target = None
813
def _unchanged(self, previous_ie):
814
"""See InventoryEntry._unchanged."""
815
compatible = super(InventoryLink, self)._unchanged(previous_ie)
816
if self.symlink_target != previous_ie.symlink_target:
820
def _snapshot_text(self, file_parents, work_tree, commit_builder):
821
"""See InventoryEntry._snapshot_text."""
822
commit_builder.modified_link(
823
self.file_id, file_parents, self.symlink_target)
826
class TreeReference(InventoryEntry):
828
kind = 'tree-reference'
830
def __init__(self, file_id, name, parent_id, revision=None,
831
reference_revision=None):
832
InventoryEntry.__init__(self, file_id, name, parent_id)
833
self.revision = revision
834
self.reference_revision = reference_revision
837
return TreeReference(self.file_id, self.name, self.parent_id,
838
self.revision, self.reference_revision)
840
def _snapshot_text(self, file_parents, work_tree, commit_builder):
841
commit_builder.modified_reference(self.file_id, file_parents)
843
def _read_tree_state(self, path, work_tree):
844
"""Populate fields in the inventory entry from the given tree.
846
self.reference_revision = work_tree.get_reference_revision(
849
def _forget_tree_state(self):
850
self.reference_revision = None
853
class Inventory(object):
854
"""Inventory of versioned files in a tree.
856
This describes which file_id is present at each point in the tree,
857
and possibly the SHA-1 or other information about the file.
858
Entries can be looked up either by path or by file_id.
860
The inventory represents a typical unix file tree, with
861
directories containing files and subdirectories. We never store
862
the full path to a file, because renaming a directory implicitly
863
moves all of its contents. This class internally maintains a
864
lookup tree that allows the children under a directory to be
867
InventoryEntry objects must not be modified after they are
868
inserted, other than through the Inventory API.
870
>>> inv = Inventory()
871
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
872
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
873
>>> inv['123-123'].name
876
May be treated as an iterator or set to look up file ids:
878
>>> bool(inv.path2id('hello.c'))
883
May also look up by name:
885
>>> [x[0] for x in inv.iter_entries()]
887
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
888
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
889
Traceback (most recent call last):
890
BzrError: parent_id {TREE_ROOT} not in inventory
891
>>> inv.add(InventoryFile('123-123', 'hello.c', 'TREE_ROOT-12345678-12345678'))
892
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT-12345678-12345678', sha1=None, len=None)
894
def __init__(self, root_id=ROOT_ID, revision_id=None):
895
"""Create or read an inventory.
897
If a working directory is specified, the inventory is read
898
from there. If the file is specified, read from that. If not,
899
the inventory is created empty.
901
The inventory is created with a default root directory, with
904
if root_id is not None:
905
assert root_id.__class__ == str
906
self._set_root(InventoryDirectory(root_id, u'', None))
910
self.revision_id = revision_id
912
def _set_root(self, ie):
914
self._byid = {self.root.file_id: self.root}
917
# TODO: jam 20051218 Should copy also copy the revision_id?
918
entries = self.iter_entries()
919
other = Inventory(entries.next()[1].file_id)
920
# copy recursively so we know directories will be added before
921
# their children. There are more efficient ways than this...
922
for path, entry in entries():
923
other.add(entry.copy())
927
return iter(self._byid)
930
"""Returns number of entries."""
931
return len(self._byid)
933
def iter_entries(self, from_dir=None):
934
"""Return (path, entry) pairs, in order by name."""
936
if self.root is None:
940
elif isinstance(from_dir, basestring):
941
from_dir = self._byid[from_dir]
943
# unrolling the recursive called changed the time from
944
# 440ms/663ms (inline/total) to 116ms/116ms
945
children = from_dir.children.items()
947
children = collections.deque(children)
948
stack = [(u'', children)]
950
from_dir_relpath, children = stack[-1]
953
name, ie = children.popleft()
955
# we know that from_dir_relpath never ends in a slash
956
# and 'f' doesn't begin with one, we can do a string op, rather
957
# than the checks of pathjoin(), though this means that all paths
959
path = from_dir_relpath + '/' + name
963
if ie.kind != 'directory':
966
# But do this child first
967
new_children = ie.children.items()
969
new_children = collections.deque(new_children)
970
stack.append((path, new_children))
971
# Break out of inner loop, so that we start outer loop with child
974
# if we finished all children, pop it off the stack
977
def iter_entries_by_dir(self, from_dir=None, specific_file_ids=None):
978
"""Iterate over the entries in a directory first order.
980
This returns all entries for a directory before returning
981
the entries for children of a directory. This is not
982
lexicographically sorted order, and is a hybrid between
983
depth-first and breadth-first.
985
:return: This yields (path, entry) pairs
987
if specific_file_ids:
988
safe = osutils.safe_file_id
989
specific_file_ids = set(safe(fid) for fid in specific_file_ids)
990
# TODO? Perhaps this should return the from_dir so that the root is
991
# yielded? or maybe an option?
993
if self.root is None:
995
# Optimize a common case
996
if specific_file_ids is not None and len(specific_file_ids) == 1:
997
file_id = list(specific_file_ids)[0]
999
yield self.id2path(file_id), self[file_id]
1001
from_dir = self.root
1002
if (specific_file_ids is None or
1003
self.root.file_id in specific_file_ids):
1004
yield u'', self.root
1005
elif isinstance(from_dir, basestring):
1006
from_dir = self._byid[from_dir]
1008
if specific_file_ids is not None:
1009
# TODO: jam 20070302 This could really be done as a loop rather
1010
# than a bunch of recursive calls.
1013
def add_ancestors(file_id):
1014
if file_id not in byid:
1016
parent_id = byid[file_id].parent_id
1017
if parent_id is None:
1019
if parent_id not in parents:
1020
parents.add(parent_id)
1021
add_ancestors(parent_id)
1022
for file_id in specific_file_ids:
1023
add_ancestors(file_id)
1027
stack = [(u'', from_dir)]
1029
cur_relpath, cur_dir = stack.pop()
1032
for child_name, child_ie in sorted(cur_dir.children.iteritems()):
1034
child_relpath = cur_relpath + child_name
1036
if (specific_file_ids is None or
1037
child_ie.file_id in specific_file_ids):
1038
yield child_relpath, child_ie
1040
if child_ie.kind == 'directory':
1041
if parents is None or child_ie.file_id in parents:
1042
child_dirs.append((child_relpath+'/', child_ie))
1043
stack.extend(reversed(child_dirs))
1046
"""Return list of (path, ie) for all entries except the root.
1048
This may be faster than iter_entries.
1051
def descend(dir_ie, dir_path):
1052
kids = dir_ie.children.items()
1054
for name, ie in kids:
1055
child_path = osutils.pathjoin(dir_path, name)
1056
accum.append((child_path, ie))
1057
if ie.kind == 'directory':
1058
descend(ie, child_path)
1060
descend(self.root, u'')
1063
def directories(self):
1064
"""Return (path, entry) pairs for all directories, including the root.
1067
def descend(parent_ie, parent_path):
1068
accum.append((parent_path, parent_ie))
1070
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
1073
for name, child_ie in kids:
1074
child_path = osutils.pathjoin(parent_path, name)
1075
descend(child_ie, child_path)
1076
descend(self.root, u'')
1079
def __contains__(self, file_id):
1080
"""True if this entry contains a file with given id.
1082
>>> inv = Inventory()
1083
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1084
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
1090
file_id = osutils.safe_file_id(file_id)
1091
return (file_id in self._byid)
1093
def __getitem__(self, file_id):
1094
"""Return the entry for given file_id.
1096
>>> inv = Inventory()
1097
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
1098
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
1099
>>> inv['123123'].name
1102
file_id = osutils.safe_file_id(file_id)
1104
return self._byid[file_id]
1106
# really we're passing an inventory, not a tree...
1107
raise errors.NoSuchId(self, file_id)
1109
def get_file_kind(self, file_id):
1110
file_id = osutils.safe_file_id(file_id)
1111
return self._byid[file_id].kind
1113
def get_child(self, parent_id, filename):
1114
parent_id = osutils.safe_file_id(parent_id)
1115
return self[parent_id].children.get(filename)
1117
def _add_child(self, entry):
1118
"""Add an entry to the inventory, without adding it to its parent"""
1119
if entry.file_id in self._byid:
1120
raise BzrError("inventory already contains entry with id {%s}" %
1122
self._byid[entry.file_id] = entry
1123
for child in getattr(entry, 'children', {}).itervalues():
1124
self._add_child(child)
1127
def add(self, entry):
1128
"""Add entry to inventory.
1130
To add a file to a branch ready to be committed, use Branch.add,
1133
Returns the new entry object.
1135
if entry.file_id in self._byid:
1136
raise errors.DuplicateFileId(entry.file_id,
1137
self._byid[entry.file_id])
1139
if entry.parent_id is None:
1140
assert self.root is None and len(self._byid) == 0
1144
parent = self._byid[entry.parent_id]
1146
raise BzrError("parent_id {%s} not in inventory" %
1149
if entry.name in parent.children:
1150
raise BzrError("%s is already versioned" %
1151
osutils.pathjoin(self.id2path(parent.file_id),
1153
parent.children[entry.name] = entry
1154
return self._add_child(entry)
1156
def add_path(self, relpath, kind, file_id=None, parent_id=None):
1157
"""Add entry from a path.
1159
The immediate parent must already be versioned.
1161
Returns the new entry object."""
1163
parts = osutils.splitpath(relpath)
1167
file_id = generate_ids.gen_root_id()
1169
file_id = osutils.safe_file_id(file_id)
1170
self.root = InventoryDirectory(file_id, '', None)
1171
self._byid = {self.root.file_id: self.root}
1174
parent_path = parts[:-1]
1175
parent_id = self.path2id(parent_path)
1176
if parent_id is None:
1177
raise errors.NotVersionedError(path=parent_path)
1178
ie = make_entry(kind, parts[-1], parent_id, file_id)
1181
def __delitem__(self, file_id):
1182
"""Remove entry by id.
1184
>>> inv = Inventory()
1185
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1186
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
1193
file_id = osutils.safe_file_id(file_id)
1196
assert ie.parent_id is None or \
1197
self[ie.parent_id].children[ie.name] == ie
1199
del self._byid[file_id]
1200
if ie.parent_id is not None:
1201
del self[ie.parent_id].children[ie.name]
1203
def __eq__(self, other):
1204
"""Compare two sets by comparing their contents.
1206
>>> i1 = Inventory()
1207
>>> i2 = Inventory()
1210
>>> i1.add(InventoryFile('123', 'foo', ROOT_ID))
1211
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
1214
>>> i2.add(InventoryFile('123', 'foo', ROOT_ID))
1215
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
1219
if not isinstance(other, Inventory):
1220
return NotImplemented
1222
return self._byid == other._byid
1224
def __ne__(self, other):
1225
return not self.__eq__(other)
1228
raise ValueError('not hashable')
1230
def _iter_file_id_parents(self, file_id):
1231
"""Yield the parents of file_id up to the root."""
1232
file_id = osutils.safe_file_id(file_id)
1233
while file_id is not None:
1235
ie = self._byid[file_id]
1237
raise errors.NoSuchId(tree=None, file_id=file_id)
1239
file_id = ie.parent_id
1241
def get_idpath(self, file_id):
1242
"""Return a list of file_ids for the path to an entry.
1244
The list contains one element for each directory followed by
1245
the id of the file itself. So the length of the returned list
1246
is equal to the depth of the file in the tree, counting the
1247
root directory as depth 1.
1249
file_id = osutils.safe_file_id(file_id)
1251
for parent in self._iter_file_id_parents(file_id):
1252
p.insert(0, parent.file_id)
1255
def id2path(self, file_id):
1256
"""Return as a string the path to file_id.
1259
>>> e = i.add(InventoryDirectory('src-id', 'src', ROOT_ID))
1260
>>> e = i.add(InventoryFile('foo-id', 'foo.c', parent_id='src-id'))
1261
>>> print i.id2path('foo-id')
1264
file_id = osutils.safe_file_id(file_id)
1265
# get all names, skipping root
1266
return '/'.join(reversed(
1267
[parent.name for parent in
1268
self._iter_file_id_parents(file_id)][:-1]))
1270
def path2id(self, name):
1271
"""Walk down through directories to return entry of last component.
1273
names may be either a list of path components, or a single
1274
string, in which case it is automatically split.
1276
This returns the entry of the last component in the path,
1277
which may be either a file or a directory.
1279
Returns None IFF the path is not found.
1281
if isinstance(name, basestring):
1282
name = osutils.splitpath(name)
1284
# mutter("lookup path %r" % name)
1291
children = getattr(parent, 'children', None)
1292
if children is None:
1295
assert cie.name == f
1296
assert cie.parent_id == parent.file_id
1299
# or raise an error?
1302
return parent.file_id
1304
def has_filename(self, names):
1305
return bool(self.path2id(names))
1307
def has_id(self, file_id):
1308
file_id = osutils.safe_file_id(file_id)
1309
return (file_id in self._byid)
1311
def remove_recursive_id(self, file_id):
1312
"""Remove file_id, and children, from the inventory.
1314
:param file_id: A file_id to remove.
1316
file_id = osutils.safe_file_id(file_id)
1317
to_find_delete = [self._byid[file_id]]
1319
while to_find_delete:
1320
ie = to_find_delete.pop()
1321
to_delete.append(ie.file_id)
1322
if ie.kind == 'directory':
1323
to_find_delete.extend(ie.children.values())
1324
for file_id in reversed(to_delete):
1326
del self._byid[file_id]
1327
if ie.parent_id is not None:
1328
del self[ie.parent_id].children[ie.name]
1330
def rename(self, file_id, new_parent_id, new_name):
1331
"""Move a file within the inventory.
1333
This can change either the name, or the parent, or both.
1335
This does not move the working file.
1337
file_id = osutils.safe_file_id(file_id)
1338
if not is_valid_name(new_name):
1339
raise BzrError("not an acceptable filename: %r" % new_name)
1341
new_parent = self._byid[new_parent_id]
1342
if new_name in new_parent.children:
1343
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
1345
new_parent_idpath = self.get_idpath(new_parent_id)
1346
if file_id in new_parent_idpath:
1347
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
1348
% (self.id2path(file_id), self.id2path(new_parent_id)))
1350
file_ie = self._byid[file_id]
1351
old_parent = self._byid[file_ie.parent_id]
1353
# TODO: Don't leave things messed up if this fails
1355
del old_parent.children[file_ie.name]
1356
new_parent.children[new_name] = file_ie
1358
file_ie.name = new_name
1359
file_ie.parent_id = new_parent_id
1361
def is_root(self, file_id):
1362
file_id = osutils.safe_file_id(file_id)
1363
return self.root is not None and file_id == self.root.file_id
1367
'directory': InventoryDirectory,
1368
'file': InventoryFile,
1369
'symlink': InventoryLink,
1370
'tree-reference': TreeReference
1373
def make_entry(kind, name, parent_id, file_id=None):
1374
"""Create an inventory entry.
1376
:param kind: the type of inventory entry to create.
1377
:param name: the basename of the entry.
1378
:param parent_id: the parent_id of the entry.
1379
:param file_id: the file_id to use. if None, one will be created.
1382
file_id = generate_ids.gen_file_id(name)
1384
file_id = osutils.safe_file_id(file_id)
1386
#------- This has been copied to bzrlib.dirstate.DirState.add, please
1387
# keep them synchronised.
1388
# we dont import normalized_filename directly because we want to be
1389
# able to change the implementation at runtime for tests.
1390
norm_name, can_access = osutils.normalized_filename(name)
1391
if norm_name != name:
1395
# TODO: jam 20060701 This would probably be more useful
1396
# if the error was raised with the full path
1397
raise errors.InvalidNormalization(name)
1400
factory = entry_factory[kind]
1402
raise BzrError("unknown kind %r" % kind)
1403
return factory(file_id, name, parent_id)
1408
def is_valid_name(name):
1410
if _NAME_RE is None:
1411
_NAME_RE = re.compile(r'^[^/\\]+$')
1413
return bool(_NAME_RE.match(name))
added
removed
bzrlib/inventory.py
