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():
115
... assert i.path2id(path)
123
>>> i.id2path('2326')
124
'src/wibble/wibble.c'
127
# Constants returned by describe_change()
129
# TODO: These should probably move to some kind of FileChangeDescription
130
# class; that's like what's inside a TreeDelta but we want to be able to
131
# generate them just for one file at a time.
133
MODIFIED_AND_RENAMED = 'modified and renamed'
137
def detect_changes(self, old_entry):
138
"""Return a (text_modified, meta_modified) from this to old_entry.
140
_read_tree_state must have been called on self and old_entry prior to
141
calling detect_changes.
145
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
146
output_to, reverse=False):
147
"""Perform a diff between two entries of the same kind."""
149
def parent_candidates(self, previous_inventories):
150
"""Find possible per-file graph parents.
152
This is currently defined by:
153
- Select the last changed revision in the parent inventory.
154
- Do deal with a short lived bug in bzr 0.8's development two entries
155
that have the same last changed but different 'x' bit settings are
158
# revision:ie mapping for each ie found in previous_inventories.
160
# identify candidate head revision ids.
161
for inv in previous_inventories:
162
if self.file_id in inv:
163
ie = inv[self.file_id]
164
assert ie.file_id == self.file_id
165
if ie.revision in candidates:
166
# same revision value in two different inventories:
167
# correct possible inconsistencies:
168
# * there was a bug in revision updates with 'x' bit
171
if candidates[ie.revision].executable != ie.executable:
172
candidates[ie.revision].executable = False
173
ie.executable = False
174
except AttributeError:
176
# must now be the same.
177
assert candidates[ie.revision] == ie
179
# add this revision as a candidate.
180
candidates[ie.revision] = ie
183
def get_tar_item(self, root, dp, now, tree):
184
"""Get a tarfile item and a file stream for its content."""
185
item = tarfile.TarInfo(osutils.pathjoin(root, dp).encode('utf8'))
186
# TODO: would be cool to actually set it to the timestamp of the
187
# revision it was last changed
189
fileobj = self._put_in_tar(item, tree)
193
"""Return true if the object this entry represents has textual data.
195
Note that textual data includes binary content.
197
Also note that all entries get weave files created for them.
198
This attribute is primarily used when upgrading from old trees that
199
did not have the weave index for all inventory entries.
203
def __init__(self, file_id, name, parent_id, text_id=None):
204
"""Create an InventoryEntry
206
The filename must be a single component, relative to the
207
parent directory; it cannot be a whole path or relative name.
209
>>> e = InventoryFile('123', 'hello.c', ROOT_ID)
214
>>> e = InventoryFile('123', 'src/hello.c', ROOT_ID)
215
Traceback (most recent call last):
216
InvalidEntryName: Invalid entry name: src/hello.c
218
assert isinstance(name, basestring), name
219
if '/' in name or '\\' in name:
220
raise errors.InvalidEntryName(name=name)
221
self.executable = False
223
self.text_sha1 = None
224
self.text_size = None
225
self.file_id = file_id
226
assert isinstance(file_id, (str, None.__class__)), \
227
'bad type %r for %r' % (type(file_id), file_id)
229
self.text_id = text_id
230
self.parent_id = parent_id
231
self.symlink_target = None
232
self.reference_revision = None
234
def kind_character(self):
235
"""Return a short kind indicator useful for appending to names."""
236
raise BzrError('unknown kind %r' % self.kind)
238
known_kinds = ('file', 'directory', 'symlink')
240
def _put_in_tar(self, item, tree):
241
"""populate item for stashing in a tar, and return the content stream.
243
If no content is available, return None.
245
raise BzrError("don't know how to export {%s} of kind %r" %
246
(self.file_id, self.kind))
248
def put_on_disk(self, dest, dp, tree):
249
"""Create a representation of self on disk in the prefix dest.
251
This is a template method - implement _put_on_disk in subclasses.
253
fullpath = osutils.pathjoin(dest, dp)
254
self._put_on_disk(fullpath, tree)
255
# mutter(" export {%s} kind %s to %s", self.file_id,
256
# self.kind, fullpath)
258
def _put_on_disk(self, fullpath, tree):
259
"""Put this entry onto disk at fullpath, from tree tree."""
260
raise BzrError("don't know how to export {%s} of kind %r" % (self.file_id, self.kind))
262
def sorted_children(self):
263
return sorted(self.children.items())
266
def versionable_kind(kind):
267
return (kind in ('file', 'directory', 'symlink', 'tree-reference'))
269
def check(self, checker, rev_id, inv, tree):
270
"""Check this inventory entry is intact.
272
This is a template method, override _check for kind specific
275
:param checker: Check object providing context for the checks;
276
can be used to find out what parts of the repository have already
278
:param rev_id: Revision id from which this InventoryEntry was loaded.
279
Not necessarily the last-changed revision for this file.
280
:param inv: Inventory from which the entry was loaded.
281
:param tree: RevisionTree for this entry.
283
if self.parent_id is not None:
284
if not inv.has_id(self.parent_id):
285
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
286
% (self.parent_id, rev_id))
287
self._check(checker, rev_id, tree)
289
def _check(self, checker, rev_id, tree):
290
"""Check this inventory entry for kind specific errors."""
291
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
295
"""Clone this inventory entry."""
296
raise NotImplementedError
299
def describe_change(old_entry, new_entry):
300
"""Describe the change between old_entry and this.
302
This smells of being an InterInventoryEntry situation, but as its
303
the first one, we're making it a static method for now.
305
An entry with a different parent, or different name is considered
306
to be renamed. Reparenting is an internal detail.
307
Note that renaming the parent does not trigger a rename for the
310
# TODO: Perhaps return an object rather than just a string
311
if old_entry is new_entry:
312
# also the case of both being None
314
elif old_entry is None:
316
elif new_entry is None:
318
if old_entry.kind != new_entry.kind:
320
text_modified, meta_modified = new_entry.detect_changes(old_entry)
321
if text_modified or meta_modified:
325
# TODO 20060511 (mbp, rbc) factor out 'detect_rename' here.
326
if old_entry.parent_id != new_entry.parent_id:
328
elif old_entry.name != new_entry.name:
332
if renamed and not modified:
333
return InventoryEntry.RENAMED
334
if modified and not renamed:
336
if modified and renamed:
337
return InventoryEntry.MODIFIED_AND_RENAMED
341
return ("%s(%r, %r, parent_id=%r, revision=%r)"
342
% (self.__class__.__name__,
348
def __eq__(self, other):
349
if not isinstance(other, InventoryEntry):
350
return NotImplemented
352
return ((self.file_id == other.file_id)
353
and (self.name == other.name)
354
and (other.symlink_target == self.symlink_target)
355
and (self.text_sha1 == other.text_sha1)
356
and (self.text_size == other.text_size)
357
and (self.text_id == other.text_id)
358
and (self.parent_id == other.parent_id)
359
and (self.kind == other.kind)
360
and (self.revision == other.revision)
361
and (self.executable == other.executable)
362
and (self.reference_revision == other.reference_revision)
365
def __ne__(self, other):
366
return not (self == other)
369
raise ValueError('not hashable')
371
def _unchanged(self, previous_ie):
372
"""Has this entry changed relative to previous_ie.
374
This method should be overridden in child classes.
377
# different inv parent
378
if previous_ie.parent_id != self.parent_id:
381
elif previous_ie.name != self.name:
383
elif previous_ie.kind != self.kind:
387
def _read_tree_state(self, path, work_tree):
388
"""Populate fields in the inventory entry from the given tree.
390
Note that this should be modified to be a noop on virtual trees
391
as all entries created there are prepopulated.
393
# TODO: Rather than running this manually, we should check the
394
# working sha1 and other expensive properties when they're
395
# first requested, or preload them if they're already known
396
pass # nothing to do by default
398
def _forget_tree_state(self):
402
class RootEntry(InventoryEntry):
404
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
405
'text_id', 'parent_id', 'children', 'executable',
406
'revision', 'symlink_target', 'reference_revision']
408
def _check(self, checker, rev_id, tree):
409
"""See InventoryEntry._check"""
411
def __init__(self, file_id):
412
self.file_id = file_id
414
self.kind = 'directory'
415
self.parent_id = None
418
symbol_versioning.warn('RootEntry is deprecated as of bzr 0.10.'
419
' Please use InventoryDirectory instead.',
420
DeprecationWarning, stacklevel=2)
422
def __eq__(self, other):
423
if not isinstance(other, RootEntry):
424
return NotImplemented
426
return (self.file_id == other.file_id) \
427
and (self.children == other.children)
430
class InventoryDirectory(InventoryEntry):
431
"""A directory in an inventory."""
433
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
434
'text_id', 'parent_id', 'children', 'executable',
435
'revision', 'symlink_target', 'reference_revision']
437
def _check(self, checker, rev_id, tree):
438
"""See InventoryEntry._check"""
439
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
440
raise BzrCheckError('directory {%s} has text in revision {%s}'
441
% (self.file_id, rev_id))
444
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
445
other.revision = self.revision
446
# note that children are *not* copied; they're pulled across when
450
def __init__(self, file_id, name, parent_id):
451
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
453
self.kind = 'directory'
455
def kind_character(self):
456
"""See InventoryEntry.kind_character."""
459
def _put_in_tar(self, item, tree):
460
"""See InventoryEntry._put_in_tar."""
461
item.type = tarfile.DIRTYPE
468
def _put_on_disk(self, fullpath, tree):
469
"""See InventoryEntry._put_on_disk."""
473
class InventoryFile(InventoryEntry):
474
"""A file in an inventory."""
476
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
477
'text_id', 'parent_id', 'children', 'executable',
478
'revision', 'symlink_target', 'reference_revision']
480
def _check(self, checker, tree_revision_id, tree):
481
"""See InventoryEntry._check"""
482
t = (self.file_id, self.revision)
483
if t in checker.checked_texts:
484
prev_sha = checker.checked_texts[t]
485
if prev_sha != self.text_sha1:
487
'mismatched sha1 on {%s} in {%s} (%s != %s) %r' %
488
(self.file_id, tree_revision_id, prev_sha, self.text_sha1,
491
checker.repeated_text_cnt += 1
494
if self.file_id not in checker.checked_weaves:
495
mutter('check weave {%s}', self.file_id)
496
w = tree._get_weave(self.file_id)
497
# Not passing a progress bar, because it creates a new
498
# progress, which overwrites the current progress,
499
# and doesn't look nice
501
checker.checked_weaves[self.file_id] = True
503
w = tree._get_weave(self.file_id)
505
mutter('check version {%s} of {%s}', tree_revision_id, self.file_id)
506
checker.checked_text_cnt += 1
507
# We can't check the length, because Weave doesn't store that
508
# information, and the whole point of looking at the weave's
509
# sha1sum is that we don't have to extract the text.
510
if self.text_sha1 != w.get_sha1s([self.revision])[0]:
511
raise BzrCheckError('text {%s} version {%s} wrong sha1'
512
% (self.file_id, self.revision))
513
checker.checked_texts[t] = self.text_sha1
516
other = InventoryFile(self.file_id, self.name, self.parent_id)
517
other.executable = self.executable
518
other.text_id = self.text_id
519
other.text_sha1 = self.text_sha1
520
other.text_size = self.text_size
521
other.revision = self.revision
524
def detect_changes(self, old_entry):
525
"""See InventoryEntry.detect_changes."""
526
assert self.text_sha1 is not None
527
assert old_entry.text_sha1 is not None
528
text_modified = (self.text_sha1 != old_entry.text_sha1)
529
meta_modified = (self.executable != old_entry.executable)
530
return text_modified, meta_modified
532
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
533
output_to, reverse=False):
534
"""See InventoryEntry._diff."""
535
from bzrlib.diff import DiffText
536
from_file_id = self.file_id
538
to_file_id = to_entry.file_id
542
to_file_id, from_file_id = from_file_id, to_file_id
543
tree, to_tree = to_tree, tree
544
from_label, to_label = to_label, from_label
545
differ = DiffText(tree, to_tree, output_to, 'utf-8', '', '',
547
return differ.diff_text(from_file_id, to_file_id, from_label, to_label)
550
"""See InventoryEntry.has_text."""
553
def __init__(self, file_id, name, parent_id):
554
super(InventoryFile, self).__init__(file_id, name, parent_id)
557
def kind_character(self):
558
"""See InventoryEntry.kind_character."""
561
def _put_in_tar(self, item, tree):
562
"""See InventoryEntry._put_in_tar."""
563
item.type = tarfile.REGTYPE
564
fileobj = tree.get_file(self.file_id)
565
item.size = self.text_size
566
if tree.is_executable(self.file_id):
572
def _put_on_disk(self, fullpath, tree):
573
"""See InventoryEntry._put_on_disk."""
574
osutils.pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
575
if tree.is_executable(self.file_id):
576
os.chmod(fullpath, 0755)
578
def _read_tree_state(self, path, work_tree):
579
"""See InventoryEntry._read_tree_state."""
580
self.text_sha1 = work_tree.get_file_sha1(self.file_id, path=path)
581
# FIXME: 20050930 probe for the text size when getting sha1
582
# in _read_tree_state
583
self.executable = work_tree.is_executable(self.file_id, path=path)
586
return ("%s(%r, %r, parent_id=%r, sha1=%r, len=%s)"
587
% (self.__class__.__name__,
594
def _forget_tree_state(self):
595
self.text_sha1 = None
597
def _unchanged(self, previous_ie):
598
"""See InventoryEntry._unchanged."""
599
compatible = super(InventoryFile, self)._unchanged(previous_ie)
600
if self.text_sha1 != previous_ie.text_sha1:
603
# FIXME: 20050930 probe for the text size when getting sha1
604
# in _read_tree_state
605
self.text_size = previous_ie.text_size
606
if self.executable != previous_ie.executable:
611
class InventoryLink(InventoryEntry):
612
"""A file in an inventory."""
614
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
615
'text_id', 'parent_id', 'children', 'executable',
616
'revision', 'symlink_target', 'reference_revision']
618
def _check(self, checker, rev_id, tree):
619
"""See InventoryEntry._check"""
620
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
621
raise BzrCheckError('symlink {%s} has text in revision {%s}'
622
% (self.file_id, rev_id))
623
if self.symlink_target is None:
624
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
625
% (self.file_id, rev_id))
628
other = InventoryLink(self.file_id, self.name, self.parent_id)
629
other.symlink_target = self.symlink_target
630
other.revision = self.revision
633
def detect_changes(self, old_entry):
634
"""See InventoryEntry.detect_changes."""
635
# FIXME: which _modified field should we use ? RBC 20051003
636
text_modified = (self.symlink_target != old_entry.symlink_target)
638
mutter(" symlink target changed")
639
meta_modified = False
640
return text_modified, meta_modified
642
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
643
output_to, reverse=False):
644
"""See InventoryEntry._diff."""
645
from bzrlib.diff import DiffSymlink
646
old_target = self.symlink_target
647
if to_entry is not None:
648
new_target = to_entry.symlink_target
657
new_target, old_target = old_target, new_target
658
differ = DiffSymlink(old_tree, new_tree, output_to)
659
return differ.diff_symlink(old_target, new_target)
661
def __init__(self, file_id, name, parent_id):
662
super(InventoryLink, self).__init__(file_id, name, parent_id)
663
self.kind = 'symlink'
665
def kind_character(self):
666
"""See InventoryEntry.kind_character."""
669
def _put_in_tar(self, item, tree):
670
"""See InventoryEntry._put_in_tar."""
671
item.type = tarfile.SYMTYPE
675
item.linkname = self.symlink_target
678
def _put_on_disk(self, fullpath, tree):
679
"""See InventoryEntry._put_on_disk."""
681
os.symlink(self.symlink_target, fullpath)
683
raise BzrError("Failed to create symlink %r -> %r, error: %s" % (fullpath, self.symlink_target, e))
685
def _read_tree_state(self, path, work_tree):
686
"""See InventoryEntry._read_tree_state."""
687
self.symlink_target = work_tree.get_symlink_target(self.file_id)
689
def _forget_tree_state(self):
690
self.symlink_target = None
692
def _unchanged(self, previous_ie):
693
"""See InventoryEntry._unchanged."""
694
compatible = super(InventoryLink, self)._unchanged(previous_ie)
695
if self.symlink_target != previous_ie.symlink_target:
700
class TreeReference(InventoryEntry):
702
kind = 'tree-reference'
704
def __init__(self, file_id, name, parent_id, revision=None,
705
reference_revision=None):
706
InventoryEntry.__init__(self, file_id, name, parent_id)
707
self.revision = revision
708
self.reference_revision = reference_revision
711
return TreeReference(self.file_id, self.name, self.parent_id,
712
self.revision, self.reference_revision)
714
def _read_tree_state(self, path, work_tree):
715
"""Populate fields in the inventory entry from the given tree.
717
self.reference_revision = work_tree.get_reference_revision(
720
def _forget_tree_state(self):
721
self.reference_revision = None
723
def _unchanged(self, previous_ie):
724
"""See InventoryEntry._unchanged."""
725
compatible = super(TreeReference, self)._unchanged(previous_ie)
726
if self.reference_revision != previous_ie.reference_revision:
731
class Inventory(object):
732
"""Inventory of versioned files in a tree.
734
This describes which file_id is present at each point in the tree,
735
and possibly the SHA-1 or other information about the file.
736
Entries can be looked up either by path or by file_id.
738
The inventory represents a typical unix file tree, with
739
directories containing files and subdirectories. We never store
740
the full path to a file, because renaming a directory implicitly
741
moves all of its contents. This class internally maintains a
742
lookup tree that allows the children under a directory to be
745
InventoryEntry objects must not be modified after they are
746
inserted, other than through the Inventory API.
748
>>> inv = Inventory()
749
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
750
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
751
>>> inv['123-123'].name
754
May be treated as an iterator or set to look up file ids:
756
>>> bool(inv.path2id('hello.c'))
761
May also look up by name:
763
>>> [x[0] for x in inv.iter_entries()]
765
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
766
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
767
Traceback (most recent call last):
768
BzrError: parent_id {TREE_ROOT} not in inventory
769
>>> inv.add(InventoryFile('123-123', 'hello.c', 'TREE_ROOT-12345678-12345678'))
770
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT-12345678-12345678', sha1=None, len=None)
772
def __init__(self, root_id=ROOT_ID, revision_id=None):
773
"""Create or read an inventory.
775
If a working directory is specified, the inventory is read
776
from there. If the file is specified, read from that. If not,
777
the inventory is created empty.
779
The inventory is created with a default root directory, with
782
if root_id is not None:
783
assert root_id.__class__ == str
784
self._set_root(InventoryDirectory(root_id, u'', None))
788
self.revision_id = revision_id
791
return "<Inventory object at %x, contents=%r>" % (id(self), self._byid)
793
def apply_delta(self, delta):
794
"""Apply a delta to this inventory.
796
:param delta: A list of changes to apply. After all the changes are
797
applied the final inventory must be internally consistent, but it
798
is ok to supply changes which, if only half-applied would have an
799
invalid result - such as supplying two changes which rename two
800
files, 'A' and 'B' with each other : [('A', 'B', 'A-id', a_entry),
801
('B', 'A', 'B-id', b_entry)].
803
Each change is a tuple, of the form (old_path, new_path, file_id,
806
When new_path is None, the change indicates the removal of an entry
807
from the inventory and new_entry will be ignored (using None is
808
appropriate). If new_path is not None, then new_entry must be an
809
InventoryEntry instance, which will be incorporated into the
810
inventory (and replace any existing entry with the same file id).
812
When old_path is None, the change indicates the addition of
813
a new entry to the inventory.
815
When neither new_path nor old_path are None, the change is a
816
modification to an entry, such as a rename, reparent, kind change
819
The children attribute of new_entry is ignored. This is because
820
this method preserves children automatically across alterations to
821
the parent of the children, and cases where the parent id of a
822
child is changing require the child to be passed in as a separate
823
change regardless. E.g. in the recursive deletion of a directory -
824
the directory's children must be included in the delta, or the
825
final inventory will be invalid.
828
# Remove all affected items which were in the original inventory,
829
# starting with the longest paths, thus ensuring parents are examined
830
# after their children, which means that everything we examine has no
831
# modified children remaining by the time we examine it.
832
for old_path, file_id in sorted(((op, f) for op, np, f, e in delta
833
if op is not None), reverse=True):
834
if file_id not in self:
837
# Preserve unaltered children of file_id for later reinsertion.
838
children[file_id] = getattr(self[file_id], 'children', {})
839
# Remove file_id and the unaltered children. If file_id is not
840
# being deleted it will be reinserted back later.
841
self.remove_recursive_id(file_id)
842
# Insert all affected which should be in the new inventory, reattaching
843
# their children if they had any. This is done from shortest path to
844
# longest, ensuring that items which were modified and whose parents in
845
# the resulting inventory were also modified, are inserted after their
847
for new_path, new_entry in sorted((np, e) for op, np, f, e in
848
delta if np is not None):
849
if new_entry.kind == 'directory':
850
new_entry.children = children.get(new_entry.file_id, {})
853
def _set_root(self, ie):
855
self._byid = {self.root.file_id: self.root}
858
# TODO: jam 20051218 Should copy also copy the revision_id?
859
entries = self.iter_entries()
860
if self.root is None:
861
return Inventory(root_id=None)
862
other = Inventory(entries.next()[1].file_id)
863
# copy recursively so we know directories will be added before
864
# their children. There are more efficient ways than this...
865
for path, entry in entries:
866
other.add(entry.copy())
870
return iter(self._byid)
873
"""Returns number of entries."""
874
return len(self._byid)
876
def iter_entries(self, from_dir=None):
877
"""Return (path, entry) pairs, in order by name."""
879
if self.root is None:
883
elif isinstance(from_dir, basestring):
884
from_dir = self._byid[from_dir]
886
# unrolling the recursive called changed the time from
887
# 440ms/663ms (inline/total) to 116ms/116ms
888
children = from_dir.children.items()
890
children = collections.deque(children)
891
stack = [(u'', children)]
893
from_dir_relpath, children = stack[-1]
896
name, ie = children.popleft()
898
# we know that from_dir_relpath never ends in a slash
899
# and 'f' doesn't begin with one, we can do a string op, rather
900
# than the checks of pathjoin(), though this means that all paths
902
path = from_dir_relpath + '/' + name
906
if ie.kind != 'directory':
909
# But do this child first
910
new_children = ie.children.items()
912
new_children = collections.deque(new_children)
913
stack.append((path, new_children))
914
# Break out of inner loop, so that we start outer loop with child
917
# if we finished all children, pop it off the stack
920
def iter_entries_by_dir(self, from_dir=None, specific_file_ids=None,
921
yield_parents=False):
922
"""Iterate over the entries in a directory first order.
924
This returns all entries for a directory before returning
925
the entries for children of a directory. This is not
926
lexicographically sorted order, and is a hybrid between
927
depth-first and breadth-first.
929
:param yield_parents: If True, yield the parents from the root leading
930
down to specific_file_ids that have been requested. This has no
931
impact if specific_file_ids is None.
932
:return: This yields (path, entry) pairs
934
if specific_file_ids and not isinstance(specific_file_ids, set):
935
specific_file_ids = set(specific_file_ids)
936
# TODO? Perhaps this should return the from_dir so that the root is
937
# yielded? or maybe an option?
939
if self.root is None:
941
# Optimize a common case
942
if (not yield_parents and specific_file_ids is not None and
943
len(specific_file_ids) == 1):
944
file_id = list(specific_file_ids)[0]
946
yield self.id2path(file_id), self[file_id]
949
if (specific_file_ids is None or yield_parents or
950
self.root.file_id in specific_file_ids):
952
elif isinstance(from_dir, basestring):
953
from_dir = self._byid[from_dir]
955
if specific_file_ids is not None:
956
# TODO: jam 20070302 This could really be done as a loop rather
957
# than a bunch of recursive calls.
960
def add_ancestors(file_id):
961
if file_id not in byid:
963
parent_id = byid[file_id].parent_id
964
if parent_id is None:
966
if parent_id not in parents:
967
parents.add(parent_id)
968
add_ancestors(parent_id)
969
for file_id in specific_file_ids:
970
add_ancestors(file_id)
974
stack = [(u'', from_dir)]
976
cur_relpath, cur_dir = stack.pop()
979
for child_name, child_ie in sorted(cur_dir.children.iteritems()):
981
child_relpath = cur_relpath + child_name
983
if (specific_file_ids is None or
984
child_ie.file_id in specific_file_ids or
985
(yield_parents and child_ie.file_id in parents)):
986
yield child_relpath, child_ie
988
if child_ie.kind == 'directory':
989
if parents is None or child_ie.file_id in parents:
990
child_dirs.append((child_relpath+'/', child_ie))
991
stack.extend(reversed(child_dirs))
993
def make_entry(self, kind, name, parent_id, file_id=None):
994
"""Simple thunk to bzrlib.inventory.make_entry."""
995
return make_entry(kind, name, parent_id, file_id)
998
"""Return list of (path, ie) for all entries except the root.
1000
This may be faster than iter_entries.
1003
def descend(dir_ie, dir_path):
1004
kids = dir_ie.children.items()
1006
for name, ie in kids:
1007
child_path = osutils.pathjoin(dir_path, name)
1008
accum.append((child_path, ie))
1009
if ie.kind == 'directory':
1010
descend(ie, child_path)
1012
descend(self.root, u'')
1015
def directories(self):
1016
"""Return (path, entry) pairs for all directories, including the root.
1019
def descend(parent_ie, parent_path):
1020
accum.append((parent_path, parent_ie))
1022
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
1025
for name, child_ie in kids:
1026
child_path = osutils.pathjoin(parent_path, name)
1027
descend(child_ie, child_path)
1028
descend(self.root, u'')
1031
def __contains__(self, file_id):
1032
"""True if this entry contains a file with given id.
1034
>>> inv = Inventory()
1035
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1036
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
1042
return (file_id in self._byid)
1044
def __getitem__(self, file_id):
1045
"""Return the entry for given file_id.
1047
>>> inv = Inventory()
1048
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
1049
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
1050
>>> inv['123123'].name
1054
return self._byid[file_id]
1056
# really we're passing an inventory, not a tree...
1057
raise errors.NoSuchId(self, file_id)
1059
def get_file_kind(self, file_id):
1060
return self._byid[file_id].kind
1062
def get_child(self, parent_id, filename):
1063
return self[parent_id].children.get(filename)
1065
def _add_child(self, entry):
1066
"""Add an entry to the inventory, without adding it to its parent"""
1067
if entry.file_id in self._byid:
1068
raise BzrError("inventory already contains entry with id {%s}" %
1070
self._byid[entry.file_id] = entry
1071
for child in getattr(entry, 'children', {}).itervalues():
1072
self._add_child(child)
1075
def add(self, entry):
1076
"""Add entry to inventory.
1078
To add a file to a branch ready to be committed, use Branch.add,
1081
Returns the new entry object.
1083
if entry.file_id in self._byid:
1084
raise errors.DuplicateFileId(entry.file_id,
1085
self._byid[entry.file_id])
1087
if entry.parent_id is None:
1088
assert self.root is None and len(self._byid) == 0
1092
parent = self._byid[entry.parent_id]
1094
raise BzrError("parent_id {%s} not in inventory" %
1097
if entry.name in parent.children:
1098
raise BzrError("%s is already versioned" %
1099
osutils.pathjoin(self.id2path(parent.file_id),
1100
entry.name).encode('utf-8'))
1101
parent.children[entry.name] = entry
1102
return self._add_child(entry)
1104
def add_path(self, relpath, kind, file_id=None, parent_id=None):
1105
"""Add entry from a path.
1107
The immediate parent must already be versioned.
1109
Returns the new entry object."""
1111
parts = osutils.splitpath(relpath)
1115
file_id = generate_ids.gen_root_id()
1116
self.root = InventoryDirectory(file_id, '', None)
1117
self._byid = {self.root.file_id: self.root}
1120
parent_path = parts[:-1]
1121
parent_id = self.path2id(parent_path)
1122
if parent_id is None:
1123
raise errors.NotVersionedError(path=parent_path)
1124
ie = make_entry(kind, parts[-1], parent_id, file_id)
1127
def __delitem__(self, file_id):
1128
"""Remove entry by id.
1130
>>> inv = Inventory()
1131
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1132
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
1141
assert ie.parent_id is None or \
1142
self[ie.parent_id].children[ie.name] == ie
1144
del self._byid[file_id]
1145
if ie.parent_id is not None:
1146
del self[ie.parent_id].children[ie.name]
1148
def __eq__(self, other):
1149
"""Compare two sets by comparing their contents.
1151
>>> i1 = Inventory()
1152
>>> i2 = Inventory()
1155
>>> i1.add(InventoryFile('123', 'foo', ROOT_ID))
1156
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
1159
>>> i2.add(InventoryFile('123', 'foo', ROOT_ID))
1160
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
1164
if not isinstance(other, Inventory):
1165
return NotImplemented
1167
return self._byid == other._byid
1169
def __ne__(self, other):
1170
return not self.__eq__(other)
1173
raise ValueError('not hashable')
1175
def _iter_file_id_parents(self, file_id):
1176
"""Yield the parents of file_id up to the root."""
1177
while file_id is not None:
1179
ie = self._byid[file_id]
1181
raise errors.NoSuchId(tree=None, file_id=file_id)
1183
file_id = ie.parent_id
1185
def get_idpath(self, file_id):
1186
"""Return a list of file_ids for the path to an entry.
1188
The list contains one element for each directory followed by
1189
the id of the file itself. So the length of the returned list
1190
is equal to the depth of the file in the tree, counting the
1191
root directory as depth 1.
1194
for parent in self._iter_file_id_parents(file_id):
1195
p.insert(0, parent.file_id)
1198
def id2path(self, file_id):
1199
"""Return as a string the path to file_id.
1202
>>> e = i.add(InventoryDirectory('src-id', 'src', ROOT_ID))
1203
>>> e = i.add(InventoryFile('foo-id', 'foo.c', parent_id='src-id'))
1204
>>> print i.id2path('foo-id')
1207
# get all names, skipping root
1208
return '/'.join(reversed(
1209
[parent.name for parent in
1210
self._iter_file_id_parents(file_id)][:-1]))
1212
def path2id(self, name):
1213
"""Walk down through directories to return entry of last component.
1215
names may be either a list of path components, or a single
1216
string, in which case it is automatically split.
1218
This returns the entry of the last component in the path,
1219
which may be either a file or a directory.
1221
Returns None IFF the path is not found.
1223
if isinstance(name, basestring):
1224
name = osutils.splitpath(name)
1226
# mutter("lookup path %r" % name)
1233
children = getattr(parent, 'children', None)
1234
if children is None:
1237
assert cie.name == f
1238
assert cie.parent_id == parent.file_id
1241
# or raise an error?
1244
return parent.file_id
1246
def has_filename(self, names):
1247
return bool(self.path2id(names))
1249
def has_id(self, file_id):
1250
return (file_id in self._byid)
1252
def remove_recursive_id(self, file_id):
1253
"""Remove file_id, and children, from the inventory.
1255
:param file_id: A file_id to remove.
1257
to_find_delete = [self._byid[file_id]]
1259
while to_find_delete:
1260
ie = to_find_delete.pop()
1261
to_delete.append(ie.file_id)
1262
if ie.kind == 'directory':
1263
to_find_delete.extend(ie.children.values())
1264
for file_id in reversed(to_delete):
1266
del self._byid[file_id]
1267
if ie.parent_id is not None:
1268
del self[ie.parent_id].children[ie.name]
1272
def rename(self, file_id, new_parent_id, new_name):
1273
"""Move a file within the inventory.
1275
This can change either the name, or the parent, or both.
1277
This does not move the working file.
1279
new_name = ensure_normalized_name(new_name)
1280
if not is_valid_name(new_name):
1281
raise BzrError("not an acceptable filename: %r" % new_name)
1283
new_parent = self._byid[new_parent_id]
1284
if new_name in new_parent.children:
1285
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
1287
new_parent_idpath = self.get_idpath(new_parent_id)
1288
if file_id in new_parent_idpath:
1289
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
1290
% (self.id2path(file_id), self.id2path(new_parent_id)))
1292
file_ie = self._byid[file_id]
1293
old_parent = self._byid[file_ie.parent_id]
1295
# TODO: Don't leave things messed up if this fails
1297
del old_parent.children[file_ie.name]
1298
new_parent.children[new_name] = file_ie
1300
file_ie.name = new_name
1301
file_ie.parent_id = new_parent_id
1303
def is_root(self, file_id):
1304
return self.root is not None and file_id == self.root.file_id
1308
'directory': InventoryDirectory,
1309
'file': InventoryFile,
1310
'symlink': InventoryLink,
1311
'tree-reference': TreeReference
1314
def make_entry(kind, name, parent_id, file_id=None):
1315
"""Create an inventory entry.
1317
:param kind: the type of inventory entry to create.
1318
:param name: the basename of the entry.
1319
:param parent_id: the parent_id of the entry.
1320
:param file_id: the file_id to use. if None, one will be created.
1323
file_id = generate_ids.gen_file_id(name)
1324
name = ensure_normalized_name(name)
1326
factory = entry_factory[kind]
1328
raise BzrError("unknown kind %r" % kind)
1329
return factory(file_id, name, parent_id)
1332
def ensure_normalized_name(name):
1335
:raises InvalidNormalization: When name is not normalized, and cannot be
1336
accessed on this platform by the normalized path.
1337
:return: The NFC normalised version of name.
1339
#------- This has been copied to bzrlib.dirstate.DirState.add, please
1340
# keep them synchronised.
1341
# we dont import normalized_filename directly because we want to be
1342
# able to change the implementation at runtime for tests.
1343
norm_name, can_access = osutils.normalized_filename(name)
1344
if norm_name != name:
1348
# TODO: jam 20060701 This would probably be more useful
1349
# if the error was raised with the full path
1350
raise errors.InvalidNormalization(name)
1356
def is_valid_name(name):
1358
if _NAME_RE is None:
1359
_NAME_RE = re.compile(r'^[^/\\]+$')
1361
return bool(_NAME_RE.match(name))
added
removed
bzrlib/inventory.py
