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.
37
from warnings import warn
40
from bzrlib import errors, osutils
41
from bzrlib.osutils import (pumpfile, quotefn, splitpath, joinpath,
42
pathjoin, sha_strings)
43
from bzrlib.errors import (NotVersionedError, InvalidEntryName,
44
BzrError, BzrCheckError, BinaryFile)
45
from bzrlib.trace import mutter
48
class InventoryEntry(object):
49
"""Description of a versioned file.
51
An InventoryEntry has the following fields, which are also
52
present in the XML inventory-entry element:
57
(within the parent directory)
60
file_id of the parent directory, or ROOT_ID
63
the revision_id in which this variation of this file was
67
Indicates that this file should be executable on systems
71
sha-1 of the text of the file
74
size in bytes of the text of the file
76
(reading a version 4 tree created a text_id field.)
81
>>> i.add(InventoryDirectory('123', 'src', ROOT_ID))
82
InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None)
83
>>> i.add(InventoryFile('2323', 'hello.c', parent_id='123'))
84
InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None)
85
>>> shouldbe = {0: '', 1: 'src', 2: pathjoin('src','hello.c')}
86
>>> for ix, j in enumerate(i.iter_entries()):
87
... print (j[0] == shouldbe[ix], j[1])
89
(True, InventoryDirectory('TREE_ROOT', '', parent_id=None, revision=None))
90
(True, InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None))
91
(True, InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None))
92
>>> i.add(InventoryFile('2323', 'bye.c', '123'))
93
Traceback (most recent call last):
95
BzrError: inventory already contains entry with id {2323}
96
>>> i.add(InventoryFile('2324', 'bye.c', '123'))
97
InventoryFile('2324', 'bye.c', parent_id='123', sha1=None, len=None)
98
>>> i.add(InventoryDirectory('2325', 'wibble', '123'))
99
InventoryDirectory('2325', 'wibble', parent_id='123', revision=None)
100
>>> i.path2id('src/wibble')
104
>>> i.add(InventoryFile('2326', 'wibble.c', '2325'))
105
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
107
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
108
>>> for path, entry in i.iter_entries():
110
... assert i.path2id(path)
118
>>> i.id2path('2326')
119
'src/wibble/wibble.c'
122
# Constants returned by describe_change()
124
# TODO: These should probably move to some kind of FileChangeDescription
125
# class; that's like what's inside a TreeDelta but we want to be able to
126
# generate them just for one file at a time.
128
MODIFIED_AND_RENAMED = 'modified and renamed'
132
def detect_changes(self, old_entry):
133
"""Return a (text_modified, meta_modified) from this to old_entry.
135
_read_tree_state must have been called on self and old_entry prior to
136
calling detect_changes.
140
def diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
141
output_to, reverse=False):
142
"""Perform a diff from this to to_entry.
144
text_diff will be used for textual difference calculation.
145
This is a template method, override _diff in child classes.
147
self._read_tree_state(tree.id2path(self.file_id), tree)
149
# cannot diff from one kind to another - you must do a removal
150
# and an addif they do not match.
151
assert self.kind == to_entry.kind
152
to_entry._read_tree_state(to_tree.id2path(to_entry.file_id),
154
self._diff(text_diff, from_label, tree, to_label, to_entry, to_tree,
157
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
158
output_to, reverse=False):
159
"""Perform a diff between two entries of the same kind."""
161
def find_previous_heads(self, previous_inventories,
162
versioned_file_store,
165
"""Return the revisions and entries that directly precede this.
167
Returned as a map from revision to inventory entry.
169
This is a map containing the file revisions in all parents
170
for which the file exists, and its revision is not a parent of
171
any other. If the file is new, the set will be empty.
173
:param versioned_file_store: A store where ancestry data on this
174
file id can be queried.
175
:param transaction: The transaction that queries to the versioned
176
file store should be completed under.
177
:param entry_vf: The entry versioned file, if its already available.
179
def get_ancestors(weave, entry):
180
return set(weave.get_ancestry(entry.revision))
181
# revision:ie mapping for each ie found in previous_inventories.
183
# revision:ie mapping with one revision for each head.
185
# revision: ancestor list for each head
187
# identify candidate head revision ids.
188
for inv in previous_inventories:
189
if self.file_id in inv:
190
ie = inv[self.file_id]
191
assert ie.file_id == self.file_id
192
if ie.revision in candidates:
193
# same revision value in two different inventories:
194
# correct possible inconsistencies:
195
# * there was a bug in revision updates with 'x' bit
198
if candidates[ie.revision].executable != ie.executable:
199
candidates[ie.revision].executable = False
200
ie.executable = False
201
except AttributeError:
203
# must now be the same.
204
assert candidates[ie.revision] == ie
206
# add this revision as a candidate.
207
candidates[ie.revision] = ie
209
# common case optimisation
210
if len(candidates) == 1:
211
# if there is only one candidate revision found
212
# then we can opening the versioned file to access ancestry:
213
# there cannot be any ancestors to eliminate when there is
214
# only one revision available.
215
heads[ie.revision] = ie
218
# eliminate ancestors amongst the available candidates:
219
# heads are those that are not an ancestor of any other candidate
220
# - this provides convergence at a per-file level.
221
for ie in candidates.values():
222
# may be an ancestor of a known head:
223
already_present = 0 != len(
224
[head for head in heads
225
if ie.revision in head_ancestors[head]])
227
# an ancestor of an analyzed candidate.
229
# not an ancestor of a known head:
230
# load the versioned file for this file id if needed
232
entry_vf = versioned_file_store.get_weave_or_empty(
233
self.file_id, transaction)
234
ancestors = get_ancestors(entry_vf, ie)
235
# may knock something else out:
236
check_heads = list(heads.keys())
237
for head in check_heads:
238
if head in ancestors:
239
# this previously discovered 'head' is not
240
# really a head - its an ancestor of the newly
243
head_ancestors[ie.revision] = ancestors
244
heads[ie.revision] = ie
247
def get_tar_item(self, root, dp, now, tree):
248
"""Get a tarfile item and a file stream for its content."""
249
item = tarfile.TarInfo(pathjoin(root, dp))
250
# TODO: would be cool to actually set it to the timestamp of the
251
# revision it was last changed
253
fileobj = self._put_in_tar(item, tree)
257
"""Return true if the object this entry represents has textual data.
259
Note that textual data includes binary content.
261
Also note that all entries get weave files created for them.
262
This attribute is primarily used when upgrading from old trees that
263
did not have the weave index for all inventory entries.
267
def __init__(self, file_id, name, parent_id, text_id=None):
268
"""Create an InventoryEntry
270
The filename must be a single component, relative to the
271
parent directory; it cannot be a whole path or relative name.
273
>>> e = InventoryFile('123', 'hello.c', ROOT_ID)
278
>>> e = InventoryFile('123', 'src/hello.c', ROOT_ID)
279
Traceback (most recent call last):
280
InvalidEntryName: Invalid entry name: src/hello.c
282
assert isinstance(name, basestring), name
283
if '/' in name or '\\' in name:
284
raise InvalidEntryName(name=name)
285
self.executable = False
287
self.text_sha1 = None
288
self.text_size = None
289
self.file_id = file_id
291
self.text_id = text_id
292
self.parent_id = parent_id
293
self.symlink_target = None
295
def kind_character(self):
296
"""Return a short kind indicator useful for appending to names."""
297
raise BzrError('unknown kind %r' % self.kind)
299
known_kinds = ('file', 'directory', 'symlink')
301
def _put_in_tar(self, item, tree):
302
"""populate item for stashing in a tar, and return the content stream.
304
If no content is available, return None.
306
raise BzrError("don't know how to export {%s} of kind %r" %
307
(self.file_id, self.kind))
309
def put_on_disk(self, dest, dp, tree):
310
"""Create a representation of self on disk in the prefix dest.
312
This is a template method - implement _put_on_disk in subclasses.
314
fullpath = pathjoin(dest, dp)
315
self._put_on_disk(fullpath, tree)
316
# mutter(" export {%s} kind %s to %s", self.file_id,
317
# self.kind, fullpath)
319
def _put_on_disk(self, fullpath, tree):
320
"""Put this entry onto disk at fullpath, from tree tree."""
321
raise BzrError("don't know how to export {%s} of kind %r" % (self.file_id, self.kind))
323
def sorted_children(self):
324
return sorted(self.children.items())
327
def versionable_kind(kind):
328
return (kind in ('file', 'directory', 'symlink'))
330
def check(self, checker, rev_id, inv, tree):
331
"""Check this inventory entry is intact.
333
This is a template method, override _check for kind specific
336
:param checker: Check object providing context for the checks;
337
can be used to find out what parts of the repository have already
339
:param rev_id: Revision id from which this InventoryEntry was loaded.
340
Not necessarily the last-changed revision for this file.
341
:param inv: Inventory from which the entry was loaded.
342
:param tree: RevisionTree for this entry.
344
if self.parent_id is not None:
345
if not inv.has_id(self.parent_id):
346
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
347
% (self.parent_id, rev_id))
348
self._check(checker, rev_id, tree)
350
def _check(self, checker, rev_id, tree):
351
"""Check this inventory entry for kind specific errors."""
352
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
356
"""Clone this inventory entry."""
357
raise NotImplementedError
360
def describe_change(old_entry, new_entry):
361
"""Describe the change between old_entry and this.
363
This smells of being an InterInventoryEntry situation, but as its
364
the first one, we're making it a static method for now.
366
An entry with a different parent, or different name is considered
367
to be renamed. Reparenting is an internal detail.
368
Note that renaming the parent does not trigger a rename for the
371
# TODO: Perhaps return an object rather than just a string
372
if old_entry is new_entry:
373
# also the case of both being None
375
elif old_entry is None:
377
elif new_entry is None:
379
text_modified, meta_modified = new_entry.detect_changes(old_entry)
380
if text_modified or meta_modified:
384
# TODO 20060511 (mbp, rbc) factor out 'detect_rename' here.
385
if old_entry.parent_id != new_entry.parent_id:
387
elif old_entry.name != new_entry.name:
391
if renamed and not modified:
392
return InventoryEntry.RENAMED
393
if modified and not renamed:
395
if modified and renamed:
396
return InventoryEntry.MODIFIED_AND_RENAMED
400
return ("%s(%r, %r, parent_id=%r, revision=%r)"
401
% (self.__class__.__name__,
407
def snapshot(self, revision, path, previous_entries,
408
work_tree, commit_builder):
409
"""Make a snapshot of this entry which may or may not have changed.
411
This means that all its fields are populated, that it has its
412
text stored in the text store or weave.
414
# mutter('new parents of %s are %r', path, previous_entries)
415
self._read_tree_state(path, work_tree)
416
# TODO: Where should we determine whether to reuse a
417
# previous revision id or create a new revision? 20060606
418
if len(previous_entries) == 1:
419
# cannot be unchanged unless there is only one parent file rev.
420
parent_ie = previous_entries.values()[0]
421
if self._unchanged(parent_ie):
422
# mutter("found unchanged entry")
423
self.revision = parent_ie.revision
425
return self._snapshot_into_revision(revision, previous_entries,
426
work_tree, commit_builder)
428
def _snapshot_into_revision(self, revision, previous_entries, work_tree,
430
"""Record this revision unconditionally into a store.
432
The entry's last-changed revision property (`revision`) is updated to
433
that of the new revision.
435
:param revision: id of the new revision that is being recorded.
437
:returns: String description of the commit (e.g. "merged", "modified"), etc.
439
# mutter('new revision {%s} for {%s}', revision, self.file_id)
440
self.revision = revision
441
self._snapshot_text(previous_entries, work_tree, commit_builder)
443
def _snapshot_text(self, file_parents, work_tree, commit_builder):
444
"""Record the 'text' of this entry, whatever form that takes.
446
This default implementation simply adds an empty text.
448
raise NotImplementedError(self._snapshot_text)
450
def __eq__(self, other):
451
if not isinstance(other, InventoryEntry):
452
return NotImplemented
454
return ((self.file_id == other.file_id)
455
and (self.name == other.name)
456
and (other.symlink_target == self.symlink_target)
457
and (self.text_sha1 == other.text_sha1)
458
and (self.text_size == other.text_size)
459
and (self.text_id == other.text_id)
460
and (self.parent_id == other.parent_id)
461
and (self.kind == other.kind)
462
and (self.revision == other.revision)
463
and (self.executable == other.executable)
466
def __ne__(self, other):
467
return not (self == other)
470
raise ValueError('not hashable')
472
def _unchanged(self, previous_ie):
473
"""Has this entry changed relative to previous_ie.
475
This method should be overridden in child classes.
478
# different inv parent
479
if previous_ie.parent_id != self.parent_id:
482
elif previous_ie.name != self.name:
486
def _read_tree_state(self, path, work_tree):
487
"""Populate fields in the inventory entry from the given tree.
489
Note that this should be modified to be a noop on virtual trees
490
as all entries created there are prepopulated.
492
# TODO: Rather than running this manually, we should check the
493
# working sha1 and other expensive properties when they're
494
# first requested, or preload them if they're already known
495
pass # nothing to do by default
497
def _forget_tree_state(self):
501
class RootEntry(InventoryEntry):
503
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
504
'text_id', 'parent_id', 'children', 'executable',
505
'revision', 'symlink_target']
507
def _check(self, checker, rev_id, tree):
508
"""See InventoryEntry._check"""
510
def __init__(self, file_id):
511
self.file_id = file_id
513
self.kind = 'directory'
514
self.parent_id = None
517
warn('RootEntry is deprecated as of bzr 0.10. Please use '
518
'InventoryDirectory instead.',
519
DeprecationWarning, stacklevel=2)
521
def __eq__(self, other):
522
if not isinstance(other, RootEntry):
523
return NotImplemented
525
return (self.file_id == other.file_id) \
526
and (self.children == other.children)
529
class InventoryDirectory(InventoryEntry):
530
"""A directory in an inventory."""
532
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
533
'text_id', 'parent_id', 'children', 'executable',
534
'revision', 'symlink_target']
536
def _check(self, checker, rev_id, tree):
537
"""See InventoryEntry._check"""
538
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
539
raise BzrCheckError('directory {%s} has text in revision {%s}'
540
% (self.file_id, rev_id))
543
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
544
other.revision = self.revision
545
# note that children are *not* copied; they're pulled across when
549
def __init__(self, file_id, name, parent_id):
550
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
552
self.kind = 'directory'
554
def kind_character(self):
555
"""See InventoryEntry.kind_character."""
558
def _put_in_tar(self, item, tree):
559
"""See InventoryEntry._put_in_tar."""
560
item.type = tarfile.DIRTYPE
567
def _put_on_disk(self, fullpath, tree):
568
"""See InventoryEntry._put_on_disk."""
571
def _snapshot_text(self, file_parents, work_tree, commit_builder):
572
"""See InventoryEntry._snapshot_text."""
573
commit_builder.modified_directory(self.file_id, file_parents)
576
class InventoryFile(InventoryEntry):
577
"""A file in an inventory."""
579
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
580
'text_id', 'parent_id', 'children', 'executable',
581
'revision', 'symlink_target']
583
def _check(self, checker, tree_revision_id, tree):
584
"""See InventoryEntry._check"""
585
t = (self.file_id, self.revision)
586
if t in checker.checked_texts:
587
prev_sha = checker.checked_texts[t]
588
if prev_sha != self.text_sha1:
589
raise BzrCheckError('mismatched sha1 on {%s} in {%s}' %
590
(self.file_id, tree_revision_id))
592
checker.repeated_text_cnt += 1
595
if self.file_id not in checker.checked_weaves:
596
mutter('check weave {%s}', self.file_id)
597
w = tree.get_weave(self.file_id)
598
# Not passing a progress bar, because it creates a new
599
# progress, which overwrites the current progress,
600
# and doesn't look nice
602
checker.checked_weaves[self.file_id] = True
604
w = tree.get_weave(self.file_id)
606
mutter('check version {%s} of {%s}', tree_revision_id, self.file_id)
607
checker.checked_text_cnt += 1
608
# We can't check the length, because Weave doesn't store that
609
# information, and the whole point of looking at the weave's
610
# sha1sum is that we don't have to extract the text.
611
if self.text_sha1 != w.get_sha1(self.revision):
612
raise BzrCheckError('text {%s} version {%s} wrong sha1'
613
% (self.file_id, self.revision))
614
checker.checked_texts[t] = self.text_sha1
617
other = InventoryFile(self.file_id, self.name, self.parent_id)
618
other.executable = self.executable
619
other.text_id = self.text_id
620
other.text_sha1 = self.text_sha1
621
other.text_size = self.text_size
622
other.revision = self.revision
625
def detect_changes(self, old_entry):
626
"""See InventoryEntry.detect_changes."""
627
assert self.text_sha1 is not None
628
assert old_entry.text_sha1 is not None
629
text_modified = (self.text_sha1 != old_entry.text_sha1)
630
meta_modified = (self.executable != old_entry.executable)
631
return text_modified, meta_modified
633
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
634
output_to, reverse=False):
635
"""See InventoryEntry._diff."""
637
from_text = tree.get_file(self.file_id).readlines()
639
to_text = to_tree.get_file(to_entry.file_id).readlines()
643
text_diff(from_label, from_text,
644
to_label, to_text, output_to)
646
text_diff(to_label, to_text,
647
from_label, from_text, output_to)
650
label_pair = (to_label, from_label)
652
label_pair = (from_label, to_label)
653
print >> output_to, "Binary files %s and %s differ" % label_pair
656
"""See InventoryEntry.has_text."""
659
def __init__(self, file_id, name, parent_id):
660
super(InventoryFile, self).__init__(file_id, name, parent_id)
663
def kind_character(self):
664
"""See InventoryEntry.kind_character."""
667
def _put_in_tar(self, item, tree):
668
"""See InventoryEntry._put_in_tar."""
669
item.type = tarfile.REGTYPE
670
fileobj = tree.get_file(self.file_id)
671
item.size = self.text_size
672
if tree.is_executable(self.file_id):
678
def _put_on_disk(self, fullpath, tree):
679
"""See InventoryEntry._put_on_disk."""
680
pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
681
if tree.is_executable(self.file_id):
682
os.chmod(fullpath, 0755)
684
def _read_tree_state(self, path, work_tree):
685
"""See InventoryEntry._read_tree_state."""
686
self.text_sha1 = work_tree.get_file_sha1(self.file_id, path=path)
687
# FIXME: 20050930 probe for the text size when getting sha1
688
# in _read_tree_state
689
self.executable = work_tree.is_executable(self.file_id, path=path)
692
return ("%s(%r, %r, parent_id=%r, sha1=%r, len=%s)"
693
% (self.__class__.__name__,
700
def _forget_tree_state(self):
701
self.text_sha1 = None
703
def _snapshot_text(self, file_parents, work_tree, commit_builder):
704
"""See InventoryEntry._snapshot_text."""
705
def get_content_byte_lines():
706
return work_tree.get_file(self.file_id).readlines()
707
self.text_sha1, self.text_size = commit_builder.modified_file_text(
708
self.file_id, file_parents, get_content_byte_lines, self.text_sha1, self.text_size)
710
def _unchanged(self, previous_ie):
711
"""See InventoryEntry._unchanged."""
712
compatible = super(InventoryFile, self)._unchanged(previous_ie)
713
if self.text_sha1 != previous_ie.text_sha1:
716
# FIXME: 20050930 probe for the text size when getting sha1
717
# in _read_tree_state
718
self.text_size = previous_ie.text_size
719
if self.executable != previous_ie.executable:
724
class InventoryLink(InventoryEntry):
725
"""A file in an inventory."""
727
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
728
'text_id', 'parent_id', 'children', 'executable',
729
'revision', 'symlink_target']
731
def _check(self, checker, rev_id, tree):
732
"""See InventoryEntry._check"""
733
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
734
raise BzrCheckError('symlink {%s} has text in revision {%s}'
735
% (self.file_id, rev_id))
736
if self.symlink_target is None:
737
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
738
% (self.file_id, rev_id))
741
other = InventoryLink(self.file_id, self.name, self.parent_id)
742
other.symlink_target = self.symlink_target
743
other.revision = self.revision
746
def detect_changes(self, old_entry):
747
"""See InventoryEntry.detect_changes."""
748
# FIXME: which _modified field should we use ? RBC 20051003
749
text_modified = (self.symlink_target != old_entry.symlink_target)
751
mutter(" symlink target changed")
752
meta_modified = False
753
return text_modified, meta_modified
755
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
756
output_to, reverse=False):
757
"""See InventoryEntry._diff."""
758
from_text = self.symlink_target
759
if to_entry is not None:
760
to_text = to_entry.symlink_target
765
print >>output_to, '=== target changed %r => %r' % (from_text, to_text)
768
print >>output_to, '=== target was %r' % self.symlink_target
770
print >>output_to, '=== target is %r' % self.symlink_target
772
def __init__(self, file_id, name, parent_id):
773
super(InventoryLink, self).__init__(file_id, name, parent_id)
774
self.kind = 'symlink'
776
def kind_character(self):
777
"""See InventoryEntry.kind_character."""
780
def _put_in_tar(self, item, tree):
781
"""See InventoryEntry._put_in_tar."""
782
item.type = tarfile.SYMTYPE
786
item.linkname = self.symlink_target
789
def _put_on_disk(self, fullpath, tree):
790
"""See InventoryEntry._put_on_disk."""
792
os.symlink(self.symlink_target, fullpath)
794
raise BzrError("Failed to create symlink %r -> %r, error: %s" % (fullpath, self.symlink_target, e))
796
def _read_tree_state(self, path, work_tree):
797
"""See InventoryEntry._read_tree_state."""
798
self.symlink_target = work_tree.get_symlink_target(self.file_id)
800
def _forget_tree_state(self):
801
self.symlink_target = None
803
def _unchanged(self, previous_ie):
804
"""See InventoryEntry._unchanged."""
805
compatible = super(InventoryLink, self)._unchanged(previous_ie)
806
if self.symlink_target != previous_ie.symlink_target:
810
def _snapshot_text(self, file_parents, work_tree, commit_builder):
811
"""See InventoryEntry._snapshot_text."""
812
commit_builder.modified_link(
813
self.file_id, file_parents, self.symlink_target)
816
class Inventory(object):
817
"""Inventory of versioned files in a tree.
819
This describes which file_id is present at each point in the tree,
820
and possibly the SHA-1 or other information about the file.
821
Entries can be looked up either by path or by file_id.
823
The inventory represents a typical unix file tree, with
824
directories containing files and subdirectories. We never store
825
the full path to a file, because renaming a directory implicitly
826
moves all of its contents. This class internally maintains a
827
lookup tree that allows the children under a directory to be
830
InventoryEntry objects must not be modified after they are
831
inserted, other than through the Inventory API.
833
>>> inv = Inventory()
834
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
835
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
836
>>> inv['123-123'].name
839
May be treated as an iterator or set to look up file ids:
841
>>> bool(inv.path2id('hello.c'))
846
May also look up by name:
848
>>> [x[0] for x in inv.iter_entries()]
850
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
851
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
852
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT-12345678-12345678', sha1=None, len=None)
854
def __init__(self, root_id=ROOT_ID, revision_id=None):
855
"""Create or read an inventory.
857
If a working directory is specified, the inventory is read
858
from there. If the file is specified, read from that. If not,
859
the inventory is created empty.
861
The inventory is created with a default root directory, with
864
# We are letting Branch.create() create a unique inventory
865
# root id. Rather than generating a random one here.
867
# root_id = bzrlib.branch.gen_file_id('TREE_ROOT')
868
if root_id is not None:
869
self._set_root(InventoryDirectory(root_id, '', None))
873
# FIXME: this isn't ever used, changing it to self.revision may break
874
# things. TODO make everything use self.revision_id
875
self.revision_id = revision_id
877
def _set_root(self, ie):
879
self._byid = {self.root.file_id: self.root}
882
# TODO: jam 20051218 Should copy also copy the revision_id?
883
entries = self.iter_entries()
884
other = Inventory(entries.next()[1].file_id)
885
# copy recursively so we know directories will be added before
886
# their children. There are more efficient ways than this...
887
for path, entry in entries():
888
other.add(entry.copy())
892
return iter(self._byid)
895
"""Returns number of entries."""
896
return len(self._byid)
898
def iter_entries(self, from_dir=None):
899
"""Return (path, entry) pairs, in order by name."""
904
elif isinstance(from_dir, basestring):
905
from_dir = self._byid[from_dir]
907
# unrolling the recursive called changed the time from
908
# 440ms/663ms (inline/total) to 116ms/116ms
909
children = from_dir.children.items()
911
children = collections.deque(children)
912
stack = [(u'', children)]
914
from_dir_relpath, children = stack[-1]
917
name, ie = children.popleft()
919
# we know that from_dir_relpath never ends in a slash
920
# and 'f' doesn't begin with one, we can do a string op, rather
921
# than the checks of pathjoin(), though this means that all paths
923
path = from_dir_relpath + '/' + name
927
if ie.kind != 'directory':
930
# But do this child first
931
new_children = ie.children.items()
933
new_children = collections.deque(new_children)
934
stack.append((path, new_children))
935
# Break out of inner loop, so that we start outer loop with child
938
# if we finished all children, pop it off the stack
941
def iter_entries_by_dir(self, from_dir=None):
942
"""Iterate over the entries in a directory first order.
944
This returns all entries for a directory before returning
945
the entries for children of a directory. This is not
946
lexicographically sorted order, and is a hybrid between
947
depth-first and breadth-first.
949
:return: This yields (path, entry) pairs
951
# TODO? Perhaps this should return the from_dir so that the root is
952
# yielded? or maybe an option?
957
elif isinstance(from_dir, basestring):
958
from_dir = self._byid[from_dir]
960
stack = [(u'', from_dir)]
962
cur_relpath, cur_dir = stack.pop()
965
for child_name, child_ie in sorted(cur_dir.children.iteritems()):
967
child_relpath = cur_relpath + child_name
969
yield child_relpath, child_ie
971
if child_ie.kind == 'directory':
972
child_dirs.append((child_relpath+'/', child_ie))
973
stack.extend(reversed(child_dirs))
976
"""Return list of (path, ie) for all entries except the root.
978
This may be faster than iter_entries.
981
def descend(dir_ie, dir_path):
982
kids = dir_ie.children.items()
984
for name, ie in kids:
985
child_path = pathjoin(dir_path, name)
986
accum.append((child_path, ie))
987
if ie.kind == 'directory':
988
descend(ie, child_path)
990
descend(self.root, u'')
993
def directories(self):
994
"""Return (path, entry) pairs for all directories, including the root.
997
def descend(parent_ie, parent_path):
998
accum.append((parent_path, parent_ie))
1000
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
1003
for name, child_ie in kids:
1004
child_path = pathjoin(parent_path, name)
1005
descend(child_ie, child_path)
1006
descend(self.root, u'')
1009
def __contains__(self, file_id):
1010
"""True if this entry contains a file with given id.
1012
>>> inv = Inventory()
1013
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1014
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
1020
return (file_id in self._byid)
1022
def __getitem__(self, file_id):
1023
"""Return the entry for given file_id.
1025
>>> inv = Inventory()
1026
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
1027
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
1028
>>> inv['123123'].name
1032
return self._byid[file_id]
1035
raise BzrError("can't look up file_id None")
1037
raise BzrError("file_id {%s} not in inventory" % file_id)
1039
def get_file_kind(self, file_id):
1040
return self._byid[file_id].kind
1042
def get_child(self, parent_id, filename):
1043
return self[parent_id].children.get(filename)
1045
def add(self, entry):
1046
"""Add entry to inventory.
1048
To add a file to a branch ready to be committed, use Branch.add,
1051
Returns the new entry object.
1053
if entry.file_id in self._byid:
1054
raise BzrError("inventory already contains entry with id {%s}" % entry.file_id)
1056
if entry.parent_id is None:
1057
assert self.root is None and len(self._byid) == 0
1058
self._set_root(entry)
1060
if entry.parent_id == ROOT_ID:
1061
assert self.root is not None, self
1062
entry.parent_id = self.root.file_id
1065
parent = self._byid[entry.parent_id]
1067
raise BzrError("parent_id {%s} not in inventory" % entry.parent_id)
1069
if entry.name in parent.children:
1070
raise BzrError("%s is already versioned" %
1071
pathjoin(self.id2path(parent.file_id), entry.name))
1073
self._byid[entry.file_id] = entry
1074
parent.children[entry.name] = entry
1077
def add_path(self, relpath, kind, file_id=None, parent_id=None):
1078
"""Add entry from a path.
1080
The immediate parent must already be versioned.
1082
Returns the new entry object."""
1084
parts = osutils.splitpath(relpath)
1088
file_id = bzrlib.workingtree.gen_root_id()
1089
self.root = InventoryDirectory(file_id, '', None)
1090
self._byid = {self.root.file_id: self.root}
1093
parent_path = parts[:-1]
1094
parent_id = self.path2id(parent_path)
1095
if parent_id is None:
1096
raise NotVersionedError(path=parent_path)
1097
ie = make_entry(kind, parts[-1], parent_id, file_id)
1100
def __delitem__(self, file_id):
1101
"""Remove entry by id.
1103
>>> inv = Inventory()
1104
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1105
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
1114
assert ie.parent_id is None or \
1115
self[ie.parent_id].children[ie.name] == ie
1117
del self._byid[file_id]
1118
if ie.parent_id is not None:
1119
del self[ie.parent_id].children[ie.name]
1121
def __eq__(self, other):
1122
"""Compare two sets by comparing their contents.
1124
>>> i1 = Inventory()
1125
>>> i2 = Inventory()
1128
>>> i1.add(InventoryFile('123', 'foo', ROOT_ID))
1129
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
1132
>>> i2.add(InventoryFile('123', 'foo', ROOT_ID))
1133
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
1137
if not isinstance(other, Inventory):
1138
return NotImplemented
1140
return self._byid == other._byid
1142
def __ne__(self, other):
1143
return not self.__eq__(other)
1146
raise ValueError('not hashable')
1148
def _iter_file_id_parents(self, file_id):
1149
"""Yield the parents of file_id up to the root."""
1150
while file_id is not None:
1152
ie = self._byid[file_id]
1154
raise BzrError("file_id {%s} not found in inventory" % file_id)
1156
file_id = ie.parent_id
1158
def get_idpath(self, file_id):
1159
"""Return a list of file_ids for the path to an entry.
1161
The list contains one element for each directory followed by
1162
the id of the file itself. So the length of the returned list
1163
is equal to the depth of the file in the tree, counting the
1164
root directory as depth 1.
1167
for parent in self._iter_file_id_parents(file_id):
1168
p.insert(0, parent.file_id)
1171
def id2path(self, file_id):
1172
"""Return as a string the path to file_id.
1175
>>> e = i.add(InventoryDirectory('src-id', 'src', ROOT_ID))
1176
>>> e = i.add(InventoryFile('foo-id', 'foo.c', parent_id='src-id'))
1177
>>> print i.id2path('foo-id')
1180
# get all names, skipping root
1181
return '/'.join(reversed(
1182
[parent.name for parent in
1183
self._iter_file_id_parents(file_id)][:-1]))
1185
def path2id(self, name):
1186
"""Walk down through directories to return entry of last component.
1188
names may be either a list of path components, or a single
1189
string, in which case it is automatically split.
1191
This returns the entry of the last component in the path,
1192
which may be either a file or a directory.
1194
Returns None IFF the path is not found.
1196
if isinstance(name, types.StringTypes):
1197
name = splitpath(name)
1199
# mutter("lookup path %r" % name)
1204
cie = parent.children[f]
1205
assert cie.name == f
1206
assert cie.parent_id == parent.file_id
1209
# or raise an error?
1212
return parent.file_id
1214
def has_filename(self, names):
1215
return bool(self.path2id(names))
1217
def has_id(self, file_id):
1218
return (file_id in self._byid)
1220
def remove_recursive_id(self, file_id):
1221
"""Remove file_id, and children, from the inventory.
1223
:param file_id: A file_id to remove.
1225
to_find_delete = [self._byid[file_id]]
1227
while to_find_delete:
1228
ie = to_find_delete.pop()
1229
to_delete.append(ie.file_id)
1230
if ie.kind == 'directory':
1231
to_find_delete.extend(ie.children.values())
1232
for file_id in reversed(to_delete):
1234
del self._byid[file_id]
1235
if ie.parent_id is not None:
1236
del self[ie.parent_id].children[ie.name]
1238
def rename(self, file_id, new_parent_id, new_name):
1239
"""Move a file within the inventory.
1241
This can change either the name, or the parent, or both.
1243
This does not move the working file."""
1244
if not is_valid_name(new_name):
1245
raise BzrError("not an acceptable filename: %r" % new_name)
1247
new_parent = self._byid[new_parent_id]
1248
if new_name in new_parent.children:
1249
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
1251
new_parent_idpath = self.get_idpath(new_parent_id)
1252
if file_id in new_parent_idpath:
1253
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
1254
% (self.id2path(file_id), self.id2path(new_parent_id)))
1256
file_ie = self._byid[file_id]
1257
old_parent = self._byid[file_ie.parent_id]
1259
# TODO: Don't leave things messed up if this fails
1261
del old_parent.children[file_ie.name]
1262
new_parent.children[new_name] = file_ie
1264
file_ie.name = new_name
1265
file_ie.parent_id = new_parent_id
1268
def make_entry(kind, name, parent_id, file_id=None):
1269
"""Create an inventory entry.
1271
:param kind: the type of inventory entry to create.
1272
:param name: the basename of the entry.
1273
:param parent_id: the parent_id of the entry.
1274
:param file_id: the file_id to use. if None, one will be created.
1277
file_id = bzrlib.workingtree.gen_file_id(name)
1279
norm_name, can_access = osutils.normalized_filename(name)
1280
if norm_name != name:
1284
# TODO: jam 20060701 This would probably be more useful
1285
# if the error was raised with the full path
1286
raise errors.InvalidNormalization(name)
1288
if kind == 'directory':
1289
return InventoryDirectory(file_id, name, parent_id)
1290
elif kind == 'file':
1291
return InventoryFile(file_id, name, parent_id)
1292
elif kind == 'symlink':
1293
return InventoryLink(file_id, name, parent_id)
1295
raise BzrError("unknown kind %r" % kind)
1300
def is_valid_name(name):
1302
if _NAME_RE is None:
1303
_NAME_RE = re.compile(r'^[^/\\]+$')
1305
return bool(_NAME_RE.match(name))
added
removed
bzrlib/inventory.py
