1
# Copyright (C) 2005, 2006, 2007 Canonical Ltd
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
# FIXME: This refactoring of the workingtree code doesn't seem to keep
18
# the WorkingTree's copy of the inventory in sync with the branch. The
19
# branch modifies its working inventory when it does a commit to make
20
# missing files permanently removed.
22
# TODO: Maybe also keep the full path of the entry, and the children?
23
# But those depend on its position within a particular inventory, and
24
# it would be nice not to need to hold the backpointer here.
26
# This should really be an id randomly assigned when the tree is
27
# created, but it's not for now.
34
from bzrlib.lazy_import import lazy_import
35
lazy_import(globals(), """
49
from bzrlib.errors import (
53
from bzrlib.trace import mutter
56
class InventoryEntry(object):
57
"""Description of a versioned file.
59
An InventoryEntry has the following fields, which are also
60
present in the XML inventory-entry element:
65
(within the parent directory)
68
file_id of the parent directory, or ROOT_ID
71
the revision_id in which this variation of this file was
75
Indicates that this file should be executable on systems
79
sha-1 of the text of the file
82
size in bytes of the text of the file
84
(reading a version 4 tree created a text_id field.)
89
>>> i.add(InventoryDirectory('123', 'src', ROOT_ID))
90
InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None)
91
>>> i.add(InventoryFile('2323', 'hello.c', parent_id='123'))
92
InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None)
93
>>> shouldbe = {0: '', 1: 'src', 2: 'src/hello.c'}
94
>>> for ix, j in enumerate(i.iter_entries()):
95
... print (j[0] == shouldbe[ix], j[1])
97
(True, InventoryDirectory('TREE_ROOT', u'', parent_id=None, revision=None))
98
(True, InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None))
99
(True, InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None))
100
>>> i.add(InventoryFile('2324', 'bye.c', '123'))
101
InventoryFile('2324', 'bye.c', parent_id='123', sha1=None, len=None)
102
>>> i.add(InventoryDirectory('2325', 'wibble', '123'))
103
InventoryDirectory('2325', 'wibble', parent_id='123', revision=None)
104
>>> i.path2id('src/wibble')
108
>>> i.add(InventoryFile('2326', 'wibble.c', '2325'))
109
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
111
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
112
>>> for path, entry in i.iter_entries():
114
... assert i.path2id(path)
122
>>> i.id2path('2326')
123
'src/wibble/wibble.c'
126
# Constants returned by describe_change()
128
# TODO: These should probably move to some kind of FileChangeDescription
129
# class; that's like what's inside a TreeDelta but we want to be able to
130
# generate them just for one file at a time.
132
MODIFIED_AND_RENAMED = 'modified and renamed'
136
def detect_changes(self, old_entry):
137
"""Return a (text_modified, meta_modified) from this to old_entry.
139
_read_tree_state must have been called on self and old_entry prior to
140
calling detect_changes.
144
def diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
145
output_to, reverse=False):
146
"""Perform a diff from this to to_entry.
148
text_diff will be used for textual difference calculation.
149
This is a template method, override _diff in child classes.
151
self._read_tree_state(tree.id2path(self.file_id), tree)
153
# cannot diff from one kind to another - you must do a removal
154
# and an addif they do not match.
155
assert self.kind == to_entry.kind
156
to_entry._read_tree_state(to_tree.id2path(to_entry.file_id),
158
self._diff(text_diff, from_label, tree, to_label, to_entry, to_tree,
161
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
162
output_to, reverse=False):
163
"""Perform a diff between two entries of the same kind."""
165
def find_previous_heads(self, previous_inventories,
166
versioned_file_store,
169
"""Return the revisions and entries that directly precede this.
171
Returned as a map from revision to inventory entry.
173
This is a map containing the file revisions in all parents
174
for which the file exists, and its revision is not a parent of
175
any other. If the file is new, the set will be empty.
177
:param versioned_file_store: A store where ancestry data on this
178
file id can be queried.
179
:param transaction: The transaction that queries to the versioned
180
file store should be completed under.
181
:param entry_vf: The entry versioned file, if its already available.
183
def get_ancestors(weave, entry):
184
return set(weave.get_ancestry(entry.revision, topo_sorted=False))
185
# revision:ie mapping for each ie found in previous_inventories.
187
# revision:ie mapping with one revision for each head.
189
# revision: ancestor list for each head
191
# identify candidate head revision ids.
192
for inv in previous_inventories:
193
if self.file_id in inv:
194
ie = inv[self.file_id]
195
assert ie.file_id == self.file_id
196
if ie.kind != self.kind:
197
# Can't be a candidate if the kind has changed.
199
if ie.revision in candidates:
200
# same revision value in two different inventories:
201
# correct possible inconsistencies:
202
# * there was a bug in revision updates with 'x' bit
205
if candidates[ie.revision].executable != ie.executable:
206
candidates[ie.revision].executable = False
207
ie.executable = False
208
except AttributeError:
210
# must now be the same.
211
assert candidates[ie.revision] == ie
213
# add this revision as a candidate.
214
candidates[ie.revision] = ie
216
# common case optimisation
217
if len(candidates) == 1:
218
# if there is only one candidate revision found
219
# then we can opening the versioned file to access ancestry:
220
# there cannot be any ancestors to eliminate when there is
221
# only one revision available.
222
heads[ie.revision] = ie
225
# eliminate ancestors amongst the available candidates:
226
# heads are those that are not an ancestor of any other candidate
227
# - this provides convergence at a per-file level.
228
for ie in candidates.values():
229
# may be an ancestor of a known head:
230
already_present = 0 != len(
231
[head for head in heads
232
if ie.revision in head_ancestors[head]])
234
# an ancestor of an analyzed candidate.
236
# not an ancestor of a known head:
237
# load the versioned file for this file id if needed
239
entry_vf = versioned_file_store.get_weave_or_empty(
240
self.file_id, transaction)
241
ancestors = get_ancestors(entry_vf, ie)
242
# may knock something else out:
243
check_heads = list(heads.keys())
244
for head in check_heads:
245
if head in ancestors:
246
# this previously discovered 'head' is not
247
# really a head - its an ancestor of the newly
250
head_ancestors[ie.revision] = ancestors
251
heads[ie.revision] = ie
254
def get_tar_item(self, root, dp, now, tree):
255
"""Get a tarfile item and a file stream for its content."""
256
item = tarfile.TarInfo(osutils.pathjoin(root, dp).encode('utf8'))
257
# TODO: would be cool to actually set it to the timestamp of the
258
# revision it was last changed
260
fileobj = self._put_in_tar(item, tree)
264
"""Return true if the object this entry represents has textual data.
266
Note that textual data includes binary content.
268
Also note that all entries get weave files created for them.
269
This attribute is primarily used when upgrading from old trees that
270
did not have the weave index for all inventory entries.
274
def __init__(self, file_id, name, parent_id, text_id=None):
275
"""Create an InventoryEntry
277
The filename must be a single component, relative to the
278
parent directory; it cannot be a whole path or relative name.
280
>>> e = InventoryFile('123', 'hello.c', ROOT_ID)
285
>>> e = InventoryFile('123', 'src/hello.c', ROOT_ID)
286
Traceback (most recent call last):
287
InvalidEntryName: Invalid entry name: src/hello.c
289
assert isinstance(name, basestring), name
290
if '/' in name or '\\' in name:
291
raise errors.InvalidEntryName(name=name)
292
self.executable = False
294
self.text_sha1 = None
295
self.text_size = None
296
self.file_id = file_id
297
assert isinstance(file_id, (str, None.__class__)), \
298
'bad type %r for %r' % (type(file_id), file_id)
300
self.text_id = text_id
301
self.parent_id = parent_id
302
self.symlink_target = None
303
self.reference_revision = None
305
def kind_character(self):
306
"""Return a short kind indicator useful for appending to names."""
307
raise BzrError('unknown kind %r' % self.kind)
309
known_kinds = ('file', 'directory', 'symlink')
311
def _put_in_tar(self, item, tree):
312
"""populate item for stashing in a tar, and return the content stream.
314
If no content is available, return None.
316
raise BzrError("don't know how to export {%s} of kind %r" %
317
(self.file_id, self.kind))
319
def put_on_disk(self, dest, dp, tree):
320
"""Create a representation of self on disk in the prefix dest.
322
This is a template method - implement _put_on_disk in subclasses.
324
fullpath = osutils.pathjoin(dest, dp)
325
self._put_on_disk(fullpath, tree)
326
# mutter(" export {%s} kind %s to %s", self.file_id,
327
# self.kind, fullpath)
329
def _put_on_disk(self, fullpath, tree):
330
"""Put this entry onto disk at fullpath, from tree tree."""
331
raise BzrError("don't know how to export {%s} of kind %r" % (self.file_id, self.kind))
333
def sorted_children(self):
334
return sorted(self.children.items())
337
def versionable_kind(kind):
338
return (kind in ('file', 'directory', 'symlink', 'tree-reference'))
340
def check(self, checker, rev_id, inv, tree):
341
"""Check this inventory entry is intact.
343
This is a template method, override _check for kind specific
346
:param checker: Check object providing context for the checks;
347
can be used to find out what parts of the repository have already
349
:param rev_id: Revision id from which this InventoryEntry was loaded.
350
Not necessarily the last-changed revision for this file.
351
:param inv: Inventory from which the entry was loaded.
352
:param tree: RevisionTree for this entry.
354
if self.parent_id is not None:
355
if not inv.has_id(self.parent_id):
356
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
357
% (self.parent_id, rev_id))
358
self._check(checker, rev_id, tree)
360
def _check(self, checker, rev_id, tree):
361
"""Check this inventory entry for kind specific errors."""
362
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
366
"""Clone this inventory entry."""
367
raise NotImplementedError
370
def describe_change(old_entry, new_entry):
371
"""Describe the change between old_entry and this.
373
This smells of being an InterInventoryEntry situation, but as its
374
the first one, we're making it a static method for now.
376
An entry with a different parent, or different name is considered
377
to be renamed. Reparenting is an internal detail.
378
Note that renaming the parent does not trigger a rename for the
381
# TODO: Perhaps return an object rather than just a string
382
if old_entry is new_entry:
383
# also the case of both being None
385
elif old_entry is None:
387
elif new_entry is None:
389
if old_entry.kind != new_entry.kind:
391
text_modified, meta_modified = new_entry.detect_changes(old_entry)
392
if text_modified or meta_modified:
396
# TODO 20060511 (mbp, rbc) factor out 'detect_rename' here.
397
if old_entry.parent_id != new_entry.parent_id:
399
elif old_entry.name != new_entry.name:
403
if renamed and not modified:
404
return InventoryEntry.RENAMED
405
if modified and not renamed:
407
if modified and renamed:
408
return InventoryEntry.MODIFIED_AND_RENAMED
412
return ("%s(%r, %r, parent_id=%r, revision=%r)"
413
% (self.__class__.__name__,
419
def snapshot(self, revision, path, previous_entries,
420
work_tree, commit_builder):
421
"""Make a snapshot of this entry which may or may not have changed.
423
This means that all its fields are populated, that it has its
424
text stored in the text store or weave.
426
:return: True if anything was recorded
428
# cannot be unchanged unless there is only one parent file rev.
429
self._read_tree_state(path, work_tree)
430
if len(previous_entries) == 1:
431
parent_ie = previous_entries.values()[0]
432
if self._unchanged(parent_ie):
433
self.revision = parent_ie.revision
435
self.revision = revision
436
return self._snapshot_text(previous_entries, work_tree, commit_builder)
438
def _snapshot_text(self, file_parents, work_tree, commit_builder):
439
"""Record the 'text' of this entry, whatever form that takes.
441
:return: True if anything was recorded
443
raise NotImplementedError(self._snapshot_text)
445
def __eq__(self, other):
446
if not isinstance(other, InventoryEntry):
447
return NotImplemented
449
return ((self.file_id == other.file_id)
450
and (self.name == other.name)
451
and (other.symlink_target == self.symlink_target)
452
and (self.text_sha1 == other.text_sha1)
453
and (self.text_size == other.text_size)
454
and (self.text_id == other.text_id)
455
and (self.parent_id == other.parent_id)
456
and (self.kind == other.kind)
457
and (self.revision == other.revision)
458
and (self.executable == other.executable)
459
and (self.reference_revision == other.reference_revision)
462
def __ne__(self, other):
463
return not (self == other)
466
raise ValueError('not hashable')
468
def _unchanged(self, previous_ie):
469
"""Has this entry changed relative to previous_ie.
471
This method should be overridden in child classes.
474
# different inv parent
475
if previous_ie.parent_id != self.parent_id:
478
elif previous_ie.name != self.name:
480
elif previous_ie.kind != self.kind:
484
def _read_tree_state(self, path, work_tree):
485
"""Populate fields in the inventory entry from the given tree.
487
Note that this should be modified to be a noop on virtual trees
488
as all entries created there are prepopulated.
490
# TODO: Rather than running this manually, we should check the
491
# working sha1 and other expensive properties when they're
492
# first requested, or preload them if they're already known
493
pass # nothing to do by default
495
def _forget_tree_state(self):
499
class RootEntry(InventoryEntry):
501
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
502
'text_id', 'parent_id', 'children', 'executable',
503
'revision', 'symlink_target', 'reference_revision']
505
def _check(self, checker, rev_id, tree):
506
"""See InventoryEntry._check"""
508
def __init__(self, file_id):
509
self.file_id = file_id
511
self.kind = 'directory'
512
self.parent_id = None
515
symbol_versioning.warn('RootEntry is deprecated as of bzr 0.10.'
516
' Please use InventoryDirectory instead.',
517
DeprecationWarning, stacklevel=2)
519
def __eq__(self, other):
520
if not isinstance(other, RootEntry):
521
return NotImplemented
523
return (self.file_id == other.file_id) \
524
and (self.children == other.children)
527
class InventoryDirectory(InventoryEntry):
528
"""A directory in an inventory."""
530
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
531
'text_id', 'parent_id', 'children', 'executable',
532
'revision', 'symlink_target', 'reference_revision']
534
def _check(self, checker, rev_id, tree):
535
"""See InventoryEntry._check"""
536
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
537
raise BzrCheckError('directory {%s} has text in revision {%s}'
538
% (self.file_id, rev_id))
541
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
542
other.revision = self.revision
543
# note that children are *not* copied; they're pulled across when
547
def __init__(self, file_id, name, parent_id):
548
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
550
self.kind = 'directory'
552
def kind_character(self):
553
"""See InventoryEntry.kind_character."""
556
def _put_in_tar(self, item, tree):
557
"""See InventoryEntry._put_in_tar."""
558
item.type = tarfile.DIRTYPE
565
def _put_on_disk(self, fullpath, tree):
566
"""See InventoryEntry._put_on_disk."""
569
def _snapshot_text(self, file_parents, work_tree, commit_builder):
570
"""See InventoryEntry._snapshot_text."""
571
commit_builder.modified_directory(self.file_id, file_parents)
575
class InventoryFile(InventoryEntry):
576
"""A file in an inventory."""
578
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
579
'text_id', 'parent_id', 'children', 'executable',
580
'revision', 'symlink_target', 'reference_revision']
582
def _check(self, checker, tree_revision_id, tree):
583
"""See InventoryEntry._check"""
584
t = (self.file_id, self.revision)
585
if t in checker.checked_texts:
586
prev_sha = checker.checked_texts[t]
587
if prev_sha != self.text_sha1:
588
raise BzrCheckError('mismatched sha1 on {%s} in {%s}' %
589
(self.file_id, tree_revision_id))
591
checker.repeated_text_cnt += 1
594
if self.file_id not in checker.checked_weaves:
595
mutter('check weave {%s}', self.file_id)
596
w = tree._get_weave(self.file_id)
597
# Not passing a progress bar, because it creates a new
598
# progress, which overwrites the current progress,
599
# and doesn't look nice
601
checker.checked_weaves[self.file_id] = True
603
w = tree._get_weave(self.file_id)
605
mutter('check version {%s} of {%s}', tree_revision_id, self.file_id)
606
checker.checked_text_cnt += 1
607
# We can't check the length, because Weave doesn't store that
608
# information, and the whole point of looking at the weave's
609
# sha1sum is that we don't have to extract the text.
610
if self.text_sha1 != w.get_sha1(self.revision):
611
raise BzrCheckError('text {%s} version {%s} wrong sha1'
612
% (self.file_id, self.revision))
613
checker.checked_texts[t] = self.text_sha1
616
other = InventoryFile(self.file_id, self.name, self.parent_id)
617
other.executable = self.executable
618
other.text_id = self.text_id
619
other.text_sha1 = self.text_sha1
620
other.text_size = self.text_size
621
other.revision = self.revision
624
def detect_changes(self, old_entry):
625
"""See InventoryEntry.detect_changes."""
626
assert self.text_sha1 is not None
627
assert old_entry.text_sha1 is not None
628
text_modified = (self.text_sha1 != old_entry.text_sha1)
629
meta_modified = (self.executable != old_entry.executable)
630
return text_modified, meta_modified
632
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
633
output_to, reverse=False):
634
"""See InventoryEntry._diff."""
636
from_text = tree.get_file(self.file_id).readlines()
638
to_text = to_tree.get_file(to_entry.file_id).readlines()
642
text_diff(from_label, from_text,
643
to_label, to_text, output_to)
645
text_diff(to_label, to_text,
646
from_label, from_text, output_to)
647
except errors.BinaryFile:
649
label_pair = (to_label, from_label)
651
label_pair = (from_label, to_label)
652
print >> output_to, \
653
("Binary files %s and %s differ" % label_pair).encode('utf8')
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
osutils.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(self, revision, path, previous_entries,
704
work_tree, commit_builder):
705
"""See InventoryEntry.snapshot."""
706
# We have a custom implementation of this for files because it's
707
# performance critical.
708
if len(previous_entries) == 0:
709
# Initial commit so we know the sha is coming later
710
self.executable = work_tree.is_executable(self.file_id, path=path)
712
self._read_tree_state(path, work_tree)
713
if len(previous_entries) == 1:
714
parent_ie = previous_entries.values()[0]
715
if self._unchanged(parent_ie):
716
self.revision = parent_ie.revision
718
self.revision = revision
719
def get_content_byte_lines():
720
return work_tree.get_file(self.file_id, path).readlines()
721
self.text_sha1, self.text_size = commit_builder.modified_file_text(
722
self.file_id, previous_entries, get_content_byte_lines,
723
self.text_sha1, self.text_size)
726
def _unchanged(self, previous_ie):
727
"""See InventoryEntry._unchanged."""
728
compatible = super(InventoryFile, self)._unchanged(previous_ie)
729
if self.text_sha1 != previous_ie.text_sha1:
732
# FIXME: 20050930 probe for the text size when getting sha1
733
# in _read_tree_state
734
self.text_size = previous_ie.text_size
735
if self.executable != previous_ie.executable:
740
class InventoryLink(InventoryEntry):
741
"""A file in an inventory."""
743
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
744
'text_id', 'parent_id', 'children', 'executable',
745
'revision', 'symlink_target', 'reference_revision']
747
def _check(self, checker, rev_id, tree):
748
"""See InventoryEntry._check"""
749
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
750
raise BzrCheckError('symlink {%s} has text in revision {%s}'
751
% (self.file_id, rev_id))
752
if self.symlink_target is None:
753
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
754
% (self.file_id, rev_id))
757
other = InventoryLink(self.file_id, self.name, self.parent_id)
758
other.symlink_target = self.symlink_target
759
other.revision = self.revision
762
def detect_changes(self, old_entry):
763
"""See InventoryEntry.detect_changes."""
764
# FIXME: which _modified field should we use ? RBC 20051003
765
text_modified = (self.symlink_target != old_entry.symlink_target)
767
mutter(" symlink target changed")
768
meta_modified = False
769
return text_modified, meta_modified
771
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
772
output_to, reverse=False):
773
"""See InventoryEntry._diff."""
774
from_text = self.symlink_target
775
if to_entry is not None:
776
to_text = to_entry.symlink_target
781
print >>output_to, '=== target changed %r => %r' % (from_text, to_text)
784
print >>output_to, '=== target was %r' % self.symlink_target
786
print >>output_to, '=== target is %r' % self.symlink_target
788
def __init__(self, file_id, name, parent_id):
789
super(InventoryLink, self).__init__(file_id, name, parent_id)
790
self.kind = 'symlink'
792
def kind_character(self):
793
"""See InventoryEntry.kind_character."""
796
def _put_in_tar(self, item, tree):
797
"""See InventoryEntry._put_in_tar."""
798
item.type = tarfile.SYMTYPE
802
item.linkname = self.symlink_target
805
def _put_on_disk(self, fullpath, tree):
806
"""See InventoryEntry._put_on_disk."""
808
os.symlink(self.symlink_target, fullpath)
810
raise BzrError("Failed to create symlink %r -> %r, error: %s" % (fullpath, self.symlink_target, e))
812
def _read_tree_state(self, path, work_tree):
813
"""See InventoryEntry._read_tree_state."""
814
self.symlink_target = work_tree.get_symlink_target(self.file_id)
816
def _forget_tree_state(self):
817
self.symlink_target = None
819
def _unchanged(self, previous_ie):
820
"""See InventoryEntry._unchanged."""
821
compatible = super(InventoryLink, self)._unchanged(previous_ie)
822
if self.symlink_target != previous_ie.symlink_target:
826
def _snapshot_text(self, file_parents, work_tree, commit_builder):
827
"""See InventoryEntry._snapshot_text."""
828
commit_builder.modified_link(
829
self.file_id, file_parents, self.symlink_target)
833
class TreeReference(InventoryEntry):
835
kind = 'tree-reference'
837
def __init__(self, file_id, name, parent_id, revision=None,
838
reference_revision=None):
839
InventoryEntry.__init__(self, file_id, name, parent_id)
840
self.revision = revision
841
self.reference_revision = reference_revision
844
return TreeReference(self.file_id, self.name, self.parent_id,
845
self.revision, self.reference_revision)
847
def _snapshot_text(self, file_parents, work_tree, commit_builder):
848
commit_builder.modified_reference(self.file_id, file_parents)
851
def _read_tree_state(self, path, work_tree):
852
"""Populate fields in the inventory entry from the given tree.
854
self.reference_revision = work_tree.get_reference_revision(
857
def _forget_tree_state(self):
858
self.reference_revision = None
861
class Inventory(object):
862
"""Inventory of versioned files in a tree.
864
This describes which file_id is present at each point in the tree,
865
and possibly the SHA-1 or other information about the file.
866
Entries can be looked up either by path or by file_id.
868
The inventory represents a typical unix file tree, with
869
directories containing files and subdirectories. We never store
870
the full path to a file, because renaming a directory implicitly
871
moves all of its contents. This class internally maintains a
872
lookup tree that allows the children under a directory to be
875
InventoryEntry objects must not be modified after they are
876
inserted, other than through the Inventory API.
878
>>> inv = Inventory()
879
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
880
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
881
>>> inv['123-123'].name
884
May be treated as an iterator or set to look up file ids:
886
>>> bool(inv.path2id('hello.c'))
891
May also look up by name:
893
>>> [x[0] for x in inv.iter_entries()]
895
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
896
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
897
Traceback (most recent call last):
898
BzrError: parent_id {TREE_ROOT} not in inventory
899
>>> inv.add(InventoryFile('123-123', 'hello.c', 'TREE_ROOT-12345678-12345678'))
900
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT-12345678-12345678', sha1=None, len=None)
902
def __init__(self, root_id=ROOT_ID, revision_id=None):
903
"""Create or read an inventory.
905
If a working directory is specified, the inventory is read
906
from there. If the file is specified, read from that. If not,
907
the inventory is created empty.
909
The inventory is created with a default root directory, with
912
if root_id is not None:
913
assert root_id.__class__ == str
914
self._set_root(InventoryDirectory(root_id, u'', None))
918
self.revision_id = revision_id
920
def _set_root(self, ie):
922
self._byid = {self.root.file_id: self.root}
925
# TODO: jam 20051218 Should copy also copy the revision_id?
926
entries = self.iter_entries()
927
other = Inventory(entries.next()[1].file_id)
928
# copy recursively so we know directories will be added before
929
# their children. There are more efficient ways than this...
930
for path, entry in entries():
931
other.add(entry.copy())
935
return iter(self._byid)
938
"""Returns number of entries."""
939
return len(self._byid)
941
def iter_entries(self, from_dir=None):
942
"""Return (path, entry) pairs, in order by name."""
944
if self.root is None:
948
elif isinstance(from_dir, basestring):
949
from_dir = self._byid[from_dir]
951
# unrolling the recursive called changed the time from
952
# 440ms/663ms (inline/total) to 116ms/116ms
953
children = from_dir.children.items()
955
children = collections.deque(children)
956
stack = [(u'', children)]
958
from_dir_relpath, children = stack[-1]
961
name, ie = children.popleft()
963
# we know that from_dir_relpath never ends in a slash
964
# and 'f' doesn't begin with one, we can do a string op, rather
965
# than the checks of pathjoin(), though this means that all paths
967
path = from_dir_relpath + '/' + name
971
if ie.kind != 'directory':
974
# But do this child first
975
new_children = ie.children.items()
977
new_children = collections.deque(new_children)
978
stack.append((path, new_children))
979
# Break out of inner loop, so that we start outer loop with child
982
# if we finished all children, pop it off the stack
985
def iter_entries_by_dir(self, from_dir=None, specific_file_ids=None):
986
"""Iterate over the entries in a directory first order.
988
This returns all entries for a directory before returning
989
the entries for children of a directory. This is not
990
lexicographically sorted order, and is a hybrid between
991
depth-first and breadth-first.
993
:return: This yields (path, entry) pairs
995
if specific_file_ids:
996
safe = osutils.safe_file_id
997
specific_file_ids = set(safe(fid) for fid in specific_file_ids)
998
# TODO? Perhaps this should return the from_dir so that the root is
999
# yielded? or maybe an option?
1000
if from_dir is None:
1001
if self.root is None:
1003
# Optimize a common case
1004
if specific_file_ids is not None and len(specific_file_ids) == 1:
1005
file_id = list(specific_file_ids)[0]
1007
yield self.id2path(file_id), self[file_id]
1009
from_dir = self.root
1010
if (specific_file_ids is None or
1011
self.root.file_id in specific_file_ids):
1012
yield u'', self.root
1013
elif isinstance(from_dir, basestring):
1014
from_dir = self._byid[from_dir]
1016
if specific_file_ids is not None:
1017
# TODO: jam 20070302 This could really be done as a loop rather
1018
# than a bunch of recursive calls.
1021
def add_ancestors(file_id):
1022
if file_id not in byid:
1024
parent_id = byid[file_id].parent_id
1025
if parent_id is None:
1027
if parent_id not in parents:
1028
parents.add(parent_id)
1029
add_ancestors(parent_id)
1030
for file_id in specific_file_ids:
1031
add_ancestors(file_id)
1035
stack = [(u'', from_dir)]
1037
cur_relpath, cur_dir = stack.pop()
1040
for child_name, child_ie in sorted(cur_dir.children.iteritems()):
1042
child_relpath = cur_relpath + child_name
1044
if (specific_file_ids is None or
1045
child_ie.file_id in specific_file_ids):
1046
yield child_relpath, child_ie
1048
if child_ie.kind == 'directory':
1049
if parents is None or child_ie.file_id in parents:
1050
child_dirs.append((child_relpath+'/', child_ie))
1051
stack.extend(reversed(child_dirs))
1053
def make_entry(self, kind, name, parent_id, file_id=None):
1054
"""Simple thunk to bzrlib.inventory.make_entry."""
1055
return make_entry(kind, name, parent_id, file_id)
1058
"""Return list of (path, ie) for all entries except the root.
1060
This may be faster than iter_entries.
1063
def descend(dir_ie, dir_path):
1064
kids = dir_ie.children.items()
1066
for name, ie in kids:
1067
child_path = osutils.pathjoin(dir_path, name)
1068
accum.append((child_path, ie))
1069
if ie.kind == 'directory':
1070
descend(ie, child_path)
1072
descend(self.root, u'')
1075
def directories(self):
1076
"""Return (path, entry) pairs for all directories, including the root.
1079
def descend(parent_ie, parent_path):
1080
accum.append((parent_path, parent_ie))
1082
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
1085
for name, child_ie in kids:
1086
child_path = osutils.pathjoin(parent_path, name)
1087
descend(child_ie, child_path)
1088
descend(self.root, u'')
1091
def __contains__(self, file_id):
1092
"""True if this entry contains a file with given id.
1094
>>> inv = Inventory()
1095
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1096
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
1102
file_id = osutils.safe_file_id(file_id)
1103
return (file_id in self._byid)
1105
def __getitem__(self, file_id):
1106
"""Return the entry for given file_id.
1108
>>> inv = Inventory()
1109
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
1110
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
1111
>>> inv['123123'].name
1114
file_id = osutils.safe_file_id(file_id)
1116
return self._byid[file_id]
1118
# really we're passing an inventory, not a tree...
1119
raise errors.NoSuchId(self, file_id)
1121
def get_file_kind(self, file_id):
1122
file_id = osutils.safe_file_id(file_id)
1123
return self._byid[file_id].kind
1125
def get_child(self, parent_id, filename):
1126
parent_id = osutils.safe_file_id(parent_id)
1127
return self[parent_id].children.get(filename)
1129
def _add_child(self, entry):
1130
"""Add an entry to the inventory, without adding it to its parent"""
1131
if entry.file_id in self._byid:
1132
raise BzrError("inventory already contains entry with id {%s}" %
1134
self._byid[entry.file_id] = entry
1135
for child in getattr(entry, 'children', {}).itervalues():
1136
self._add_child(child)
1139
def add(self, entry):
1140
"""Add entry to inventory.
1142
To add a file to a branch ready to be committed, use Branch.add,
1145
Returns the new entry object.
1147
if entry.file_id in self._byid:
1148
raise errors.DuplicateFileId(entry.file_id,
1149
self._byid[entry.file_id])
1151
if entry.parent_id is None:
1152
assert self.root is None and len(self._byid) == 0
1156
parent = self._byid[entry.parent_id]
1158
raise BzrError("parent_id {%s} not in inventory" %
1161
if entry.name in parent.children:
1162
raise BzrError("%s is already versioned" %
1163
osutils.pathjoin(self.id2path(parent.file_id),
1164
entry.name).encode('utf-8'))
1165
parent.children[entry.name] = entry
1166
return self._add_child(entry)
1168
def add_path(self, relpath, kind, file_id=None, parent_id=None):
1169
"""Add entry from a path.
1171
The immediate parent must already be versioned.
1173
Returns the new entry object."""
1175
parts = osutils.splitpath(relpath)
1179
file_id = generate_ids.gen_root_id()
1181
file_id = osutils.safe_file_id(file_id)
1182
self.root = InventoryDirectory(file_id, '', None)
1183
self._byid = {self.root.file_id: self.root}
1186
parent_path = parts[:-1]
1187
parent_id = self.path2id(parent_path)
1188
if parent_id is None:
1189
raise errors.NotVersionedError(path=parent_path)
1190
ie = make_entry(kind, parts[-1], parent_id, file_id)
1193
def __delitem__(self, file_id):
1194
"""Remove entry by id.
1196
>>> inv = Inventory()
1197
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1198
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
1205
file_id = osutils.safe_file_id(file_id)
1208
assert ie.parent_id is None or \
1209
self[ie.parent_id].children[ie.name] == ie
1211
del self._byid[file_id]
1212
if ie.parent_id is not None:
1213
del self[ie.parent_id].children[ie.name]
1215
def __eq__(self, other):
1216
"""Compare two sets by comparing their contents.
1218
>>> i1 = Inventory()
1219
>>> i2 = Inventory()
1222
>>> i1.add(InventoryFile('123', 'foo', ROOT_ID))
1223
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
1226
>>> i2.add(InventoryFile('123', 'foo', ROOT_ID))
1227
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
1231
if not isinstance(other, Inventory):
1232
return NotImplemented
1234
return self._byid == other._byid
1236
def __ne__(self, other):
1237
return not self.__eq__(other)
1240
raise ValueError('not hashable')
1242
def _iter_file_id_parents(self, file_id):
1243
"""Yield the parents of file_id up to the root."""
1244
file_id = osutils.safe_file_id(file_id)
1245
while file_id is not None:
1247
ie = self._byid[file_id]
1249
raise errors.NoSuchId(tree=None, file_id=file_id)
1251
file_id = ie.parent_id
1253
def get_idpath(self, file_id):
1254
"""Return a list of file_ids for the path to an entry.
1256
The list contains one element for each directory followed by
1257
the id of the file itself. So the length of the returned list
1258
is equal to the depth of the file in the tree, counting the
1259
root directory as depth 1.
1261
file_id = osutils.safe_file_id(file_id)
1263
for parent in self._iter_file_id_parents(file_id):
1264
p.insert(0, parent.file_id)
1267
def id2path(self, file_id):
1268
"""Return as a string the path to file_id.
1271
>>> e = i.add(InventoryDirectory('src-id', 'src', ROOT_ID))
1272
>>> e = i.add(InventoryFile('foo-id', 'foo.c', parent_id='src-id'))
1273
>>> print i.id2path('foo-id')
1276
file_id = osutils.safe_file_id(file_id)
1277
# get all names, skipping root
1278
return '/'.join(reversed(
1279
[parent.name for parent in
1280
self._iter_file_id_parents(file_id)][:-1]))
1282
def path2id(self, name):
1283
"""Walk down through directories to return entry of last component.
1285
names may be either a list of path components, or a single
1286
string, in which case it is automatically split.
1288
This returns the entry of the last component in the path,
1289
which may be either a file or a directory.
1291
Returns None IFF the path is not found.
1293
if isinstance(name, basestring):
1294
name = osutils.splitpath(name)
1296
# mutter("lookup path %r" % name)
1303
children = getattr(parent, 'children', None)
1304
if children is None:
1307
assert cie.name == f
1308
assert cie.parent_id == parent.file_id
1311
# or raise an error?
1314
return parent.file_id
1316
def has_filename(self, names):
1317
return bool(self.path2id(names))
1319
def has_id(self, file_id):
1320
file_id = osutils.safe_file_id(file_id)
1321
return (file_id in self._byid)
1323
def remove_recursive_id(self, file_id):
1324
"""Remove file_id, and children, from the inventory.
1326
:param file_id: A file_id to remove.
1328
file_id = osutils.safe_file_id(file_id)
1329
to_find_delete = [self._byid[file_id]]
1331
while to_find_delete:
1332
ie = to_find_delete.pop()
1333
to_delete.append(ie.file_id)
1334
if ie.kind == 'directory':
1335
to_find_delete.extend(ie.children.values())
1336
for file_id in reversed(to_delete):
1338
del self._byid[file_id]
1339
if ie.parent_id is not None:
1340
del self[ie.parent_id].children[ie.name]
1344
def rename(self, file_id, new_parent_id, new_name):
1345
"""Move a file within the inventory.
1347
This can change either the name, or the parent, or both.
1349
This does not move the working file.
1351
file_id = osutils.safe_file_id(file_id)
1352
if not is_valid_name(new_name):
1353
raise BzrError("not an acceptable filename: %r" % new_name)
1355
new_parent = self._byid[new_parent_id]
1356
if new_name in new_parent.children:
1357
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
1359
new_parent_idpath = self.get_idpath(new_parent_id)
1360
if file_id in new_parent_idpath:
1361
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
1362
% (self.id2path(file_id), self.id2path(new_parent_id)))
1364
file_ie = self._byid[file_id]
1365
old_parent = self._byid[file_ie.parent_id]
1367
# TODO: Don't leave things messed up if this fails
1369
del old_parent.children[file_ie.name]
1370
new_parent.children[new_name] = file_ie
1372
file_ie.name = new_name
1373
file_ie.parent_id = new_parent_id
1375
def is_root(self, file_id):
1376
file_id = osutils.safe_file_id(file_id)
1377
return self.root is not None and file_id == self.root.file_id
1381
'directory': InventoryDirectory,
1382
'file': InventoryFile,
1383
'symlink': InventoryLink,
1384
'tree-reference': TreeReference
1387
def make_entry(kind, name, parent_id, file_id=None):
1388
"""Create an inventory entry.
1390
:param kind: the type of inventory entry to create.
1391
:param name: the basename of the entry.
1392
:param parent_id: the parent_id of the entry.
1393
:param file_id: the file_id to use. if None, one will be created.
1396
file_id = generate_ids.gen_file_id(name)
1398
file_id = osutils.safe_file_id(file_id)
1400
#------- This has been copied to bzrlib.dirstate.DirState.add, please
1401
# keep them synchronised.
1402
# we dont import normalized_filename directly because we want to be
1403
# able to change the implementation at runtime for tests.
1404
norm_name, can_access = osutils.normalized_filename(name)
1405
if norm_name != name:
1409
# TODO: jam 20060701 This would probably be more useful
1410
# if the error was raised with the full path
1411
raise errors.InvalidNormalization(name)
1414
factory = entry_factory[kind]
1416
raise BzrError("unknown kind %r" % kind)
1417
return factory(file_id, name, parent_id)
1422
def is_valid_name(name):
1424
if _NAME_RE is None:
1425
_NAME_RE = re.compile(r'^[^/\\]+$')
1427
return bool(_NAME_RE.match(name))
added
removed
bzrlib/inventory.py
