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.symbol_versioning import deprecated_method
54
from bzrlib.trace import mutter
57
class InventoryEntry(object):
58
"""Description of a versioned file.
60
An InventoryEntry has the following fields, which are also
61
present in the XML inventory-entry element:
66
(within the parent directory)
69
file_id of the parent directory, or ROOT_ID
72
the revision_id in which this variation of this file was
76
Indicates that this file should be executable on systems
80
sha-1 of the text of the file
83
size in bytes of the text of the file
85
(reading a version 4 tree created a text_id field.)
90
>>> i.add(InventoryDirectory('123', 'src', ROOT_ID))
91
InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None)
92
>>> i.add(InventoryFile('2323', 'hello.c', parent_id='123'))
93
InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None)
94
>>> shouldbe = {0: '', 1: 'src', 2: 'src/hello.c'}
95
>>> for ix, j in enumerate(i.iter_entries()):
96
... print (j[0] == shouldbe[ix], j[1])
98
(True, InventoryDirectory('TREE_ROOT', u'', parent_id=None, revision=None))
99
(True, InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None))
100
(True, InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None))
101
>>> i.add(InventoryFile('2324', 'bye.c', '123'))
102
InventoryFile('2324', 'bye.c', parent_id='123', sha1=None, len=None)
103
>>> i.add(InventoryDirectory('2325', 'wibble', '123'))
104
InventoryDirectory('2325', 'wibble', parent_id='123', revision=None)
105
>>> i.path2id('src/wibble')
109
>>> i.add(InventoryFile('2326', 'wibble.c', '2325'))
110
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
112
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
113
>>> for path, entry in i.iter_entries():
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 between two entries of the same kind."""
148
def parent_candidates(self, previous_inventories):
149
"""Find possible per-file graph parents.
151
This is currently defined by:
152
- Select the last changed revision in the parent inventory.
153
- Do deal with a short lived bug in bzr 0.8's development two entries
154
that have the same last changed but different 'x' bit settings are
157
# revision:ie mapping for each ie found in previous_inventories.
159
# identify candidate head revision ids.
160
for inv in previous_inventories:
161
if self.file_id in inv:
162
ie = inv[self.file_id]
163
if ie.revision in candidates:
164
# same revision value in two different inventories:
165
# correct possible inconsistencies:
166
# * there was a bug in revision updates with 'x' bit
169
if candidates[ie.revision].executable != ie.executable:
170
candidates[ie.revision].executable = False
171
ie.executable = False
172
except AttributeError:
175
# add this revision as a candidate.
176
candidates[ie.revision] = ie
179
def get_tar_item(self, root, dp, now, tree):
180
"""Get a tarfile item and a file stream for its content."""
181
item = tarfile.TarInfo(osutils.pathjoin(root, dp).encode('utf8'))
182
# TODO: would be cool to actually set it to the timestamp of the
183
# revision it was last changed
185
fileobj = self._put_in_tar(item, tree)
189
"""Return true if the object this entry represents has textual data.
191
Note that textual data includes binary content.
193
Also note that all entries get weave files created for them.
194
This attribute is primarily used when upgrading from old trees that
195
did not have the weave index for all inventory entries.
199
def __init__(self, file_id, name, parent_id, text_id=None):
200
"""Create an InventoryEntry
202
The filename must be a single component, relative to the
203
parent directory; it cannot be a whole path or relative name.
205
>>> e = InventoryFile('123', 'hello.c', ROOT_ID)
210
>>> e = InventoryFile('123', 'src/hello.c', ROOT_ID)
211
Traceback (most recent call last):
212
InvalidEntryName: Invalid entry name: src/hello.c
214
if '/' in name or '\\' in name:
215
raise errors.InvalidEntryName(name=name)
216
self.executable = False
218
self.text_sha1 = None
219
self.text_size = None
220
self.file_id = file_id
222
self.text_id = text_id
223
self.parent_id = parent_id
224
self.symlink_target = None
225
self.reference_revision = None
227
def kind_character(self):
228
"""Return a short kind indicator useful for appending to names."""
229
raise BzrError('unknown kind %r' % self.kind)
231
known_kinds = ('file', 'directory', 'symlink')
233
def _put_in_tar(self, item, tree):
234
"""populate item for stashing in a tar, and return the content stream.
236
If no content is available, return None.
238
raise BzrError("don't know how to export {%s} of kind %r" %
239
(self.file_id, self.kind))
241
def put_on_disk(self, dest, dp, tree):
242
"""Create a representation of self on disk in the prefix dest.
244
This is a template method - implement _put_on_disk in subclasses.
246
fullpath = osutils.pathjoin(dest, dp)
247
self._put_on_disk(fullpath, tree)
248
# mutter(" export {%s} kind %s to %s", self.file_id,
249
# self.kind, fullpath)
251
def _put_on_disk(self, fullpath, tree):
252
"""Put this entry onto disk at fullpath, from tree tree."""
253
raise BzrError("don't know how to export {%s} of kind %r" % (self.file_id, self.kind))
255
def sorted_children(self):
256
return sorted(self.children.items())
259
def versionable_kind(kind):
260
return (kind in ('file', 'directory', 'symlink', 'tree-reference'))
262
def check(self, checker, rev_id, inv, tree):
263
"""Check this inventory entry is intact.
265
This is a template method, override _check for kind specific
268
:param checker: Check object providing context for the checks;
269
can be used to find out what parts of the repository have already
271
:param rev_id: Revision id from which this InventoryEntry was loaded.
272
Not necessarily the last-changed revision for this file.
273
:param inv: Inventory from which the entry was loaded.
274
:param tree: RevisionTree for this entry.
276
if self.parent_id is not None:
277
if not inv.has_id(self.parent_id):
278
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
279
% (self.parent_id, rev_id))
280
self._check(checker, rev_id, tree)
282
def _check(self, checker, rev_id, tree):
283
"""Check this inventory entry for kind specific errors."""
284
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
288
"""Clone this inventory entry."""
289
raise NotImplementedError
292
def describe_change(old_entry, new_entry):
293
"""Describe the change between old_entry and this.
295
This smells of being an InterInventoryEntry situation, but as its
296
the first one, we're making it a static method for now.
298
An entry with a different parent, or different name is considered
299
to be renamed. Reparenting is an internal detail.
300
Note that renaming the parent does not trigger a rename for the
303
# TODO: Perhaps return an object rather than just a string
304
if old_entry is new_entry:
305
# also the case of both being None
307
elif old_entry is None:
309
elif new_entry is None:
311
if old_entry.kind != new_entry.kind:
313
text_modified, meta_modified = new_entry.detect_changes(old_entry)
314
if text_modified or meta_modified:
318
# TODO 20060511 (mbp, rbc) factor out 'detect_rename' here.
319
if old_entry.parent_id != new_entry.parent_id:
321
elif old_entry.name != new_entry.name:
325
if renamed and not modified:
326
return InventoryEntry.RENAMED
327
if modified and not renamed:
329
if modified and renamed:
330
return InventoryEntry.MODIFIED_AND_RENAMED
334
return ("%s(%r, %r, parent_id=%r, revision=%r)"
335
% (self.__class__.__name__,
341
def __eq__(self, other):
342
if not isinstance(other, InventoryEntry):
343
return NotImplemented
345
return ((self.file_id == other.file_id)
346
and (self.name == other.name)
347
and (other.symlink_target == self.symlink_target)
348
and (self.text_sha1 == other.text_sha1)
349
and (self.text_size == other.text_size)
350
and (self.text_id == other.text_id)
351
and (self.parent_id == other.parent_id)
352
and (self.kind == other.kind)
353
and (self.revision == other.revision)
354
and (self.executable == other.executable)
355
and (self.reference_revision == other.reference_revision)
358
def __ne__(self, other):
359
return not (self == other)
362
raise ValueError('not hashable')
364
def _unchanged(self, previous_ie):
365
"""Has this entry changed relative to previous_ie.
367
This method should be overridden in child classes.
370
# different inv parent
371
if previous_ie.parent_id != self.parent_id:
374
elif previous_ie.name != self.name:
376
elif previous_ie.kind != self.kind:
380
def _read_tree_state(self, path, work_tree):
381
"""Populate fields in the inventory entry from the given tree.
383
Note that this should be modified to be a noop on virtual trees
384
as all entries created there are prepopulated.
386
# TODO: Rather than running this manually, we should check the
387
# working sha1 and other expensive properties when they're
388
# first requested, or preload them if they're already known
389
pass # nothing to do by default
391
def _forget_tree_state(self):
395
class RootEntry(InventoryEntry):
397
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
398
'text_id', 'parent_id', 'children', 'executable',
399
'revision', 'symlink_target', 'reference_revision']
401
def _check(self, checker, rev_id, tree):
402
"""See InventoryEntry._check"""
404
def __init__(self, file_id):
405
self.file_id = file_id
407
self.kind = 'directory'
408
self.parent_id = None
411
symbol_versioning.warn('RootEntry is deprecated as of bzr 0.10.'
412
' Please use InventoryDirectory instead.',
413
DeprecationWarning, stacklevel=2)
415
def __eq__(self, other):
416
if not isinstance(other, RootEntry):
417
return NotImplemented
419
return (self.file_id == other.file_id) \
420
and (self.children == other.children)
423
class InventoryDirectory(InventoryEntry):
424
"""A directory in an inventory."""
426
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
427
'text_id', 'parent_id', 'children', 'executable',
428
'revision', 'symlink_target', 'reference_revision']
430
def _check(self, checker, rev_id, tree):
431
"""See InventoryEntry._check"""
432
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
433
raise BzrCheckError('directory {%s} has text in revision {%s}'
434
% (self.file_id, rev_id))
437
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
438
other.revision = self.revision
439
# note that children are *not* copied; they're pulled across when
443
def __init__(self, file_id, name, parent_id):
444
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
446
self.kind = 'directory'
448
def kind_character(self):
449
"""See InventoryEntry.kind_character."""
452
def _put_in_tar(self, item, tree):
453
"""See InventoryEntry._put_in_tar."""
454
item.type = tarfile.DIRTYPE
461
def _put_on_disk(self, fullpath, tree):
462
"""See InventoryEntry._put_on_disk."""
466
class InventoryFile(InventoryEntry):
467
"""A file in an inventory."""
469
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
470
'text_id', 'parent_id', 'children', 'executable',
471
'revision', 'symlink_target', 'reference_revision']
473
def _check(self, checker, tree_revision_id, tree):
474
"""See InventoryEntry._check"""
475
t = (self.file_id, self.revision)
476
if t in checker.checked_texts:
477
prev_sha = checker.checked_texts[t]
478
if prev_sha != self.text_sha1:
480
'mismatched sha1 on {%s} in {%s} (%s != %s) %r' %
481
(self.file_id, tree_revision_id, prev_sha, self.text_sha1,
484
checker.repeated_text_cnt += 1
487
if self.file_id not in checker.checked_weaves:
488
mutter('check weave {%s}', self.file_id)
489
w = tree._get_weave(self.file_id)
490
# Not passing a progress bar, because it creates a new
491
# progress, which overwrites the current progress,
492
# and doesn't look nice
494
checker.checked_weaves[self.file_id] = True
496
w = tree._get_weave(self.file_id)
498
mutter('check version {%s} of {%s}', tree_revision_id, self.file_id)
499
checker.checked_text_cnt += 1
500
# We can't check the length, because Weave doesn't store that
501
# information, and the whole point of looking at the weave's
502
# sha1sum is that we don't have to extract the text.
503
if self.text_sha1 != w.get_sha1s([self.revision])[0]:
504
raise BzrCheckError('text {%s} version {%s} wrong sha1'
505
% (self.file_id, self.revision))
506
checker.checked_texts[t] = self.text_sha1
509
other = InventoryFile(self.file_id, self.name, self.parent_id)
510
other.executable = self.executable
511
other.text_id = self.text_id
512
other.text_sha1 = self.text_sha1
513
other.text_size = self.text_size
514
other.revision = self.revision
517
def detect_changes(self, old_entry):
518
"""See InventoryEntry.detect_changes."""
519
text_modified = (self.text_sha1 != old_entry.text_sha1)
520
meta_modified = (self.executable != old_entry.executable)
521
return text_modified, meta_modified
523
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
524
output_to, reverse=False):
525
"""See InventoryEntry._diff."""
526
from bzrlib.diff import DiffText
527
from_file_id = self.file_id
529
to_file_id = to_entry.file_id
533
to_file_id, from_file_id = from_file_id, to_file_id
534
tree, to_tree = to_tree, tree
535
from_label, to_label = to_label, from_label
536
differ = DiffText(tree, to_tree, output_to, 'utf-8', '', '',
538
return differ.diff_text(from_file_id, to_file_id, from_label, to_label)
541
"""See InventoryEntry.has_text."""
544
def __init__(self, file_id, name, parent_id):
545
super(InventoryFile, self).__init__(file_id, name, parent_id)
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.REGTYPE
555
fileobj = tree.get_file(self.file_id)
556
item.size = self.text_size
557
if tree.is_executable(self.file_id):
563
def _put_on_disk(self, fullpath, tree):
564
"""See InventoryEntry._put_on_disk."""
565
osutils.pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
566
if tree.is_executable(self.file_id):
567
os.chmod(fullpath, 0755)
569
def _read_tree_state(self, path, work_tree):
570
"""See InventoryEntry._read_tree_state."""
571
self.text_sha1 = work_tree.get_file_sha1(self.file_id, path=path)
572
# FIXME: 20050930 probe for the text size when getting sha1
573
# in _read_tree_state
574
self.executable = work_tree.is_executable(self.file_id, path=path)
577
return ("%s(%r, %r, parent_id=%r, sha1=%r, len=%s)"
578
% (self.__class__.__name__,
585
def _forget_tree_state(self):
586
self.text_sha1 = None
588
def _unchanged(self, previous_ie):
589
"""See InventoryEntry._unchanged."""
590
compatible = super(InventoryFile, self)._unchanged(previous_ie)
591
if self.text_sha1 != previous_ie.text_sha1:
594
# FIXME: 20050930 probe for the text size when getting sha1
595
# in _read_tree_state
596
self.text_size = previous_ie.text_size
597
if self.executable != previous_ie.executable:
602
class InventoryLink(InventoryEntry):
603
"""A file in an inventory."""
605
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
606
'text_id', 'parent_id', 'children', 'executable',
607
'revision', 'symlink_target', 'reference_revision']
609
def _check(self, checker, rev_id, tree):
610
"""See InventoryEntry._check"""
611
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
612
raise BzrCheckError('symlink {%s} has text in revision {%s}'
613
% (self.file_id, rev_id))
614
if self.symlink_target is None:
615
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
616
% (self.file_id, rev_id))
619
other = InventoryLink(self.file_id, self.name, self.parent_id)
620
other.symlink_target = self.symlink_target
621
other.revision = self.revision
624
def detect_changes(self, old_entry):
625
"""See InventoryEntry.detect_changes."""
626
# FIXME: which _modified field should we use ? RBC 20051003
627
text_modified = (self.symlink_target != old_entry.symlink_target)
629
mutter(" symlink target changed")
630
meta_modified = False
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."""
636
from bzrlib.diff import DiffSymlink
637
old_target = self.symlink_target
638
if to_entry is not None:
639
new_target = to_entry.symlink_target
648
new_target, old_target = old_target, new_target
649
differ = DiffSymlink(old_tree, new_tree, output_to)
650
return differ.diff_symlink(old_target, new_target)
652
def __init__(self, file_id, name, parent_id):
653
super(InventoryLink, self).__init__(file_id, name, parent_id)
654
self.kind = 'symlink'
656
def kind_character(self):
657
"""See InventoryEntry.kind_character."""
660
def _put_in_tar(self, item, tree):
661
"""See InventoryEntry._put_in_tar."""
662
item.type = tarfile.SYMTYPE
666
item.linkname = self.symlink_target
669
def _put_on_disk(self, fullpath, tree):
670
"""See InventoryEntry._put_on_disk."""
672
os.symlink(self.symlink_target, fullpath)
674
raise BzrError("Failed to create symlink %r -> %r, error: %s" % (fullpath, self.symlink_target, e))
676
def _read_tree_state(self, path, work_tree):
677
"""See InventoryEntry._read_tree_state."""
678
self.symlink_target = work_tree.get_symlink_target(self.file_id)
680
def _forget_tree_state(self):
681
self.symlink_target = None
683
def _unchanged(self, previous_ie):
684
"""See InventoryEntry._unchanged."""
685
compatible = super(InventoryLink, self)._unchanged(previous_ie)
686
if self.symlink_target != previous_ie.symlink_target:
691
class TreeReference(InventoryEntry):
693
kind = 'tree-reference'
695
def __init__(self, file_id, name, parent_id, revision=None,
696
reference_revision=None):
697
InventoryEntry.__init__(self, file_id, name, parent_id)
698
self.revision = revision
699
self.reference_revision = reference_revision
702
return TreeReference(self.file_id, self.name, self.parent_id,
703
self.revision, self.reference_revision)
705
def _read_tree_state(self, path, work_tree):
706
"""Populate fields in the inventory entry from the given tree.
708
self.reference_revision = work_tree.get_reference_revision(
711
def _forget_tree_state(self):
712
self.reference_revision = None
714
def _unchanged(self, previous_ie):
715
"""See InventoryEntry._unchanged."""
716
compatible = super(TreeReference, self)._unchanged(previous_ie)
717
if self.reference_revision != previous_ie.reference_revision:
722
class Inventory(object):
723
"""Inventory of versioned files in a tree.
725
This describes which file_id is present at each point in the tree,
726
and possibly the SHA-1 or other information about the file.
727
Entries can be looked up either by path or by file_id.
729
The inventory represents a typical unix file tree, with
730
directories containing files and subdirectories. We never store
731
the full path to a file, because renaming a directory implicitly
732
moves all of its contents. This class internally maintains a
733
lookup tree that allows the children under a directory to be
736
InventoryEntry objects must not be modified after they are
737
inserted, other than through the Inventory API.
739
>>> inv = Inventory()
740
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
741
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
742
>>> inv['123-123'].name
745
May be treated as an iterator or set to look up file ids:
747
>>> bool(inv.path2id('hello.c'))
752
May also look up by name:
754
>>> [x[0] for x in inv.iter_entries()]
756
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
757
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
758
Traceback (most recent call last):
759
BzrError: parent_id {TREE_ROOT} not in inventory
760
>>> inv.add(InventoryFile('123-123', 'hello.c', 'TREE_ROOT-12345678-12345678'))
761
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT-12345678-12345678', sha1=None, len=None)
763
def __init__(self, root_id=ROOT_ID, revision_id=None):
764
"""Create or read an inventory.
766
If a working directory is specified, the inventory is read
767
from there. If the file is specified, read from that. If not,
768
the inventory is created empty.
770
The inventory is created with a default root directory, with
773
if root_id is not None:
774
self._set_root(InventoryDirectory(root_id, u'', None))
778
self.revision_id = revision_id
781
return "<Inventory object at %x, contents=%r>" % (id(self), self._byid)
783
def apply_delta(self, delta):
784
"""Apply a delta to this inventory.
786
:param delta: A list of changes to apply. After all the changes are
787
applied the final inventory must be internally consistent, but it
788
is ok to supply changes which, if only half-applied would have an
789
invalid result - such as supplying two changes which rename two
790
files, 'A' and 'B' with each other : [('A', 'B', 'A-id', a_entry),
791
('B', 'A', 'B-id', b_entry)].
793
Each change is a tuple, of the form (old_path, new_path, file_id,
796
When new_path is None, the change indicates the removal of an entry
797
from the inventory and new_entry will be ignored (using None is
798
appropriate). If new_path is not None, then new_entry must be an
799
InventoryEntry instance, which will be incorporated into the
800
inventory (and replace any existing entry with the same file id).
802
When old_path is None, the change indicates the addition of
803
a new entry to the inventory.
805
When neither new_path nor old_path are None, the change is a
806
modification to an entry, such as a rename, reparent, kind change
809
The children attribute of new_entry is ignored. This is because
810
this method preserves children automatically across alterations to
811
the parent of the children, and cases where the parent id of a
812
child is changing require the child to be passed in as a separate
813
change regardless. E.g. in the recursive deletion of a directory -
814
the directory's children must be included in the delta, or the
815
final inventory will be invalid.
818
# Remove all affected items which were in the original inventory,
819
# starting with the longest paths, thus ensuring parents are examined
820
# after their children, which means that everything we examine has no
821
# modified children remaining by the time we examine it.
822
for old_path, file_id in sorted(((op, f) for op, np, f, e in delta
823
if op is not None), reverse=True):
824
if file_id not in self:
827
# Preserve unaltered children of file_id for later reinsertion.
828
children[file_id] = getattr(self[file_id], 'children', {})
829
# Remove file_id and the unaltered children. If file_id is not
830
# being deleted it will be reinserted back later.
831
self.remove_recursive_id(file_id)
832
# Insert all affected which should be in the new inventory, reattaching
833
# their children if they had any. This is done from shortest path to
834
# longest, ensuring that items which were modified and whose parents in
835
# the resulting inventory were also modified, are inserted after their
837
for new_path, new_entry in sorted((np, e) for op, np, f, e in
838
delta if np is not None):
839
if new_entry.kind == 'directory':
840
new_entry.children = children.get(new_entry.file_id, {})
843
def _set_root(self, ie):
845
self._byid = {self.root.file_id: self.root}
848
# TODO: jam 20051218 Should copy also copy the revision_id?
849
entries = self.iter_entries()
850
if self.root is None:
851
return Inventory(root_id=None)
852
other = Inventory(entries.next()[1].file_id)
853
# copy recursively so we know directories will be added before
854
# their children. There are more efficient ways than this...
855
for path, entry in entries:
856
other.add(entry.copy())
860
return iter(self._byid)
863
"""Returns number of entries."""
864
return len(self._byid)
866
def iter_entries(self, from_dir=None):
867
"""Return (path, entry) pairs, in order by name."""
869
if self.root is None:
873
elif isinstance(from_dir, basestring):
874
from_dir = self._byid[from_dir]
876
# unrolling the recursive called changed the time from
877
# 440ms/663ms (inline/total) to 116ms/116ms
878
children = from_dir.children.items()
880
children = collections.deque(children)
881
stack = [(u'', children)]
883
from_dir_relpath, children = stack[-1]
886
name, ie = children.popleft()
888
# we know that from_dir_relpath never ends in a slash
889
# and 'f' doesn't begin with one, we can do a string op, rather
890
# than the checks of pathjoin(), though this means that all paths
892
path = from_dir_relpath + '/' + name
896
if ie.kind != 'directory':
899
# But do this child first
900
new_children = ie.children.items()
902
new_children = collections.deque(new_children)
903
stack.append((path, new_children))
904
# Break out of inner loop, so that we start outer loop with child
907
# if we finished all children, pop it off the stack
910
def iter_entries_by_dir(self, from_dir=None, specific_file_ids=None,
911
yield_parents=False):
912
"""Iterate over the entries in a directory first order.
914
This returns all entries for a directory before returning
915
the entries for children of a directory. This is not
916
lexicographically sorted order, and is a hybrid between
917
depth-first and breadth-first.
919
:param yield_parents: If True, yield the parents from the root leading
920
down to specific_file_ids that have been requested. This has no
921
impact if specific_file_ids is None.
922
:return: This yields (path, entry) pairs
924
if specific_file_ids and not isinstance(specific_file_ids, set):
925
specific_file_ids = set(specific_file_ids)
926
# TODO? Perhaps this should return the from_dir so that the root is
927
# yielded? or maybe an option?
929
if self.root is None:
931
# Optimize a common case
932
if (not yield_parents and specific_file_ids is not None and
933
len(specific_file_ids) == 1):
934
file_id = list(specific_file_ids)[0]
936
yield self.id2path(file_id), self[file_id]
939
if (specific_file_ids is None or yield_parents or
940
self.root.file_id in specific_file_ids):
942
elif isinstance(from_dir, basestring):
943
from_dir = self._byid[from_dir]
945
if specific_file_ids is not None:
946
# TODO: jam 20070302 This could really be done as a loop rather
947
# than a bunch of recursive calls.
950
def add_ancestors(file_id):
951
if file_id not in byid:
953
parent_id = byid[file_id].parent_id
954
if parent_id is None:
956
if parent_id not in parents:
957
parents.add(parent_id)
958
add_ancestors(parent_id)
959
for file_id in specific_file_ids:
960
add_ancestors(file_id)
964
stack = [(u'', from_dir)]
966
cur_relpath, cur_dir = stack.pop()
969
for child_name, child_ie in sorted(cur_dir.children.iteritems()):
971
child_relpath = cur_relpath + child_name
973
if (specific_file_ids is None or
974
child_ie.file_id in specific_file_ids or
975
(yield_parents and child_ie.file_id in parents)):
976
yield child_relpath, child_ie
978
if child_ie.kind == 'directory':
979
if parents is None or child_ie.file_id in parents:
980
child_dirs.append((child_relpath+'/', child_ie))
981
stack.extend(reversed(child_dirs))
983
def make_entry(self, kind, name, parent_id, file_id=None):
984
"""Simple thunk to bzrlib.inventory.make_entry."""
985
return make_entry(kind, name, parent_id, file_id)
988
"""Return list of (path, ie) for all entries except the root.
990
This may be faster than iter_entries.
993
def descend(dir_ie, dir_path):
994
kids = dir_ie.children.items()
996
for name, ie in kids:
997
child_path = osutils.pathjoin(dir_path, name)
998
accum.append((child_path, ie))
999
if ie.kind == 'directory':
1000
descend(ie, child_path)
1002
descend(self.root, u'')
1005
def directories(self):
1006
"""Return (path, entry) pairs for all directories, including the root.
1009
def descend(parent_ie, parent_path):
1010
accum.append((parent_path, parent_ie))
1012
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
1015
for name, child_ie in kids:
1016
child_path = osutils.pathjoin(parent_path, name)
1017
descend(child_ie, child_path)
1018
descend(self.root, u'')
1021
def __contains__(self, file_id):
1022
"""True if this entry contains a file with given id.
1024
>>> inv = Inventory()
1025
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1026
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
1032
return (file_id in self._byid)
1034
def __getitem__(self, file_id):
1035
"""Return the entry for given file_id.
1037
>>> inv = Inventory()
1038
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
1039
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
1040
>>> inv['123123'].name
1044
return self._byid[file_id]
1046
# really we're passing an inventory, not a tree...
1047
raise errors.NoSuchId(self, file_id)
1049
def get_file_kind(self, file_id):
1050
return self._byid[file_id].kind
1052
def get_child(self, parent_id, filename):
1053
return self[parent_id].children.get(filename)
1055
def _add_child(self, entry):
1056
"""Add an entry to the inventory, without adding it to its parent"""
1057
if entry.file_id in self._byid:
1058
raise BzrError("inventory already contains entry with id {%s}" %
1060
self._byid[entry.file_id] = entry
1061
for child in getattr(entry, 'children', {}).itervalues():
1062
self._add_child(child)
1065
def add(self, entry):
1066
"""Add entry to inventory.
1068
To add a file to a branch ready to be committed, use Branch.add,
1071
Returns the new entry object.
1073
if entry.file_id in self._byid:
1074
raise errors.DuplicateFileId(entry.file_id,
1075
self._byid[entry.file_id])
1077
if entry.parent_id is None:
1081
parent = self._byid[entry.parent_id]
1083
raise BzrError("parent_id {%s} not in inventory" %
1086
if entry.name in parent.children:
1087
raise BzrError("%s is already versioned" %
1088
osutils.pathjoin(self.id2path(parent.file_id),
1089
entry.name).encode('utf-8'))
1090
parent.children[entry.name] = entry
1091
return self._add_child(entry)
1093
def add_path(self, relpath, kind, file_id=None, parent_id=None):
1094
"""Add entry from a path.
1096
The immediate parent must already be versioned.
1098
Returns the new entry object."""
1100
parts = osutils.splitpath(relpath)
1104
file_id = generate_ids.gen_root_id()
1105
self.root = InventoryDirectory(file_id, '', None)
1106
self._byid = {self.root.file_id: self.root}
1109
parent_path = parts[:-1]
1110
parent_id = self.path2id(parent_path)
1111
if parent_id is None:
1112
raise errors.NotVersionedError(path=parent_path)
1113
ie = make_entry(kind, parts[-1], parent_id, file_id)
1116
def __delitem__(self, file_id):
1117
"""Remove entry by id.
1119
>>> inv = Inventory()
1120
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1121
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
1129
del self._byid[file_id]
1130
if ie.parent_id is not None:
1131
del self[ie.parent_id].children[ie.name]
1133
def __eq__(self, other):
1134
"""Compare two sets by comparing their contents.
1136
>>> i1 = Inventory()
1137
>>> i2 = Inventory()
1140
>>> i1.add(InventoryFile('123', 'foo', ROOT_ID))
1141
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
1144
>>> i2.add(InventoryFile('123', 'foo', ROOT_ID))
1145
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
1149
if not isinstance(other, Inventory):
1150
return NotImplemented
1152
return self._byid == other._byid
1154
def __ne__(self, other):
1155
return not self.__eq__(other)
1158
raise ValueError('not hashable')
1160
def _iter_file_id_parents(self, file_id):
1161
"""Yield the parents of file_id up to the root."""
1162
while file_id is not None:
1164
ie = self._byid[file_id]
1166
raise errors.NoSuchId(tree=None, file_id=file_id)
1168
file_id = ie.parent_id
1170
def get_idpath(self, file_id):
1171
"""Return a list of file_ids for the path to an entry.
1173
The list contains one element for each directory followed by
1174
the id of the file itself. So the length of the returned list
1175
is equal to the depth of the file in the tree, counting the
1176
root directory as depth 1.
1179
for parent in self._iter_file_id_parents(file_id):
1180
p.insert(0, parent.file_id)
1183
def id2path(self, file_id):
1184
"""Return as a string the path to file_id.
1187
>>> e = i.add(InventoryDirectory('src-id', 'src', ROOT_ID))
1188
>>> e = i.add(InventoryFile('foo-id', 'foo.c', parent_id='src-id'))
1189
>>> print i.id2path('foo-id')
1192
# get all names, skipping root
1193
return '/'.join(reversed(
1194
[parent.name for parent in
1195
self._iter_file_id_parents(file_id)][:-1]))
1197
def path2id(self, name):
1198
"""Walk down through directories to return entry of last component.
1200
names may be either a list of path components, or a single
1201
string, in which case it is automatically split.
1203
This returns the entry of the last component in the path,
1204
which may be either a file or a directory.
1206
Returns None IFF the path is not found.
1208
if isinstance(name, basestring):
1209
name = osutils.splitpath(name)
1211
# mutter("lookup path %r" % name)
1218
children = getattr(parent, 'children', None)
1219
if children is None:
1224
# or raise an error?
1227
return parent.file_id
1229
def has_filename(self, names):
1230
return bool(self.path2id(names))
1232
def has_id(self, file_id):
1233
return (file_id in self._byid)
1235
def remove_recursive_id(self, file_id):
1236
"""Remove file_id, and children, from the inventory.
1238
:param file_id: A file_id to remove.
1240
to_find_delete = [self._byid[file_id]]
1242
while to_find_delete:
1243
ie = to_find_delete.pop()
1244
to_delete.append(ie.file_id)
1245
if ie.kind == 'directory':
1246
to_find_delete.extend(ie.children.values())
1247
for file_id in reversed(to_delete):
1249
del self._byid[file_id]
1250
if ie.parent_id is not None:
1251
del self[ie.parent_id].children[ie.name]
1255
def rename(self, file_id, new_parent_id, new_name):
1256
"""Move a file within the inventory.
1258
This can change either the name, or the parent, or both.
1260
This does not move the working file.
1262
new_name = ensure_normalized_name(new_name)
1263
if not is_valid_name(new_name):
1264
raise BzrError("not an acceptable filename: %r" % new_name)
1266
new_parent = self._byid[new_parent_id]
1267
if new_name in new_parent.children:
1268
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
1270
new_parent_idpath = self.get_idpath(new_parent_id)
1271
if file_id in new_parent_idpath:
1272
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
1273
% (self.id2path(file_id), self.id2path(new_parent_id)))
1275
file_ie = self._byid[file_id]
1276
old_parent = self._byid[file_ie.parent_id]
1278
# TODO: Don't leave things messed up if this fails
1280
del old_parent.children[file_ie.name]
1281
new_parent.children[new_name] = file_ie
1283
file_ie.name = new_name
1284
file_ie.parent_id = new_parent_id
1286
def is_root(self, file_id):
1287
return self.root is not None and file_id == self.root.file_id
1291
'directory': InventoryDirectory,
1292
'file': InventoryFile,
1293
'symlink': InventoryLink,
1294
'tree-reference': TreeReference
1297
def make_entry(kind, name, parent_id, file_id=None):
1298
"""Create an inventory entry.
1300
:param kind: the type of inventory entry to create.
1301
:param name: the basename of the entry.
1302
:param parent_id: the parent_id of the entry.
1303
:param file_id: the file_id to use. if None, one will be created.
1306
file_id = generate_ids.gen_file_id(name)
1307
name = ensure_normalized_name(name)
1309
factory = entry_factory[kind]
1311
raise BzrError("unknown kind %r" % kind)
1312
return factory(file_id, name, parent_id)
1315
def ensure_normalized_name(name):
1318
:raises InvalidNormalization: When name is not normalized, and cannot be
1319
accessed on this platform by the normalized path.
1320
:return: The NFC normalised version of name.
1322
#------- This has been copied to bzrlib.dirstate.DirState.add, please
1323
# keep them synchronised.
1324
# we dont import normalized_filename directly because we want to be
1325
# able to change the implementation at runtime for tests.
1326
norm_name, can_access = osutils.normalized_filename(name)
1327
if norm_name != name:
1331
# TODO: jam 20060701 This would probably be more useful
1332
# if the error was raised with the full path
1333
raise errors.InvalidNormalization(name)
1339
def is_valid_name(name):
1341
if _NAME_RE is None:
1342
_NAME_RE = re.compile(r'^[^/\\]+$')
1344
return bool(_NAME_RE.match(name))
added
removed
bzrlib/inventory.py
