1
# Copyright (C) 2005, 2006 Canonical Ltd
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
# FIXME: This refactoring of the workingtree code doesn't seem to keep
18
# the WorkingTree's copy of the inventory in sync with the branch. The
19
# branch modifies its working inventory when it does a commit to make
20
# missing files permanently removed.
22
# TODO: Maybe also keep the full path of the entry, and the children?
23
# But those depend on its position within a particular inventory, and
24
# it would be nice not to need to hold the backpointer here.
26
# This should really be an id randomly assigned when the tree is
27
# created, but it's not for now.
39
from bzrlib import errors, osutils
40
from bzrlib.osutils import (pumpfile, quotefn, splitpath, joinpath,
41
pathjoin, sha_strings)
42
from bzrlib.errors import (NotVersionedError, InvalidEntryName,
43
BzrError, BzrCheckError, BinaryFile)
44
from bzrlib.trace import mutter
47
class InventoryEntry(object):
48
"""Description of a versioned file.
50
An InventoryEntry has the following fields, which are also
51
present in the XML inventory-entry element:
56
(within the parent directory)
59
file_id of the parent directory, or ROOT_ID
62
the revision_id in which this variation of this file was
66
Indicates that this file should be executable on systems
70
sha-1 of the text of the file
73
size in bytes of the text of the file
75
(reading a version 4 tree created a text_id field.)
80
>>> i.add(InventoryDirectory('123', 'src', ROOT_ID))
81
InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None)
82
>>> i.add(InventoryFile('2323', 'hello.c', parent_id='123'))
83
InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None)
84
>>> shouldbe = {0: '', 1: 'src', 2: pathjoin('src','hello.c')}
85
>>> for ix, j in enumerate(i.iter_entries()):
86
... print (j[0] == shouldbe[ix], j[1])
88
(True, InventoryDirectory('TREE_ROOT', '', parent_id=None, revision=None))
89
(True, InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None))
90
(True, InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None))
91
>>> i.add(InventoryFile('2323', 'bye.c', '123'))
92
Traceback (most recent call last):
94
BzrError: inventory already contains entry with id {2323}
95
>>> i.add(InventoryFile('2324', 'bye.c', '123'))
96
InventoryFile('2324', 'bye.c', parent_id='123', sha1=None, len=None)
97
>>> i.add(InventoryDirectory('2325', 'wibble', '123'))
98
InventoryDirectory('2325', 'wibble', parent_id='123', revision=None)
99
>>> i.path2id('src/wibble')
103
>>> i.add(InventoryFile('2326', 'wibble.c', '2325'))
104
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
106
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
107
>>> for path, entry in i.iter_entries():
109
... assert i.path2id(path)
117
>>> i.id2path('2326')
118
'src/wibble/wibble.c'
121
# Constants returned by describe_change()
123
# TODO: These should probably move to some kind of FileChangeDescription
124
# class; that's like what's inside a TreeDelta but we want to be able to
125
# generate them just for one file at a time.
127
MODIFIED_AND_RENAMED = 'modified and renamed'
131
def detect_changes(self, old_entry):
132
"""Return a (text_modified, meta_modified) from this to old_entry.
134
_read_tree_state must have been called on self and old_entry prior to
135
calling detect_changes.
139
def diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
140
output_to, reverse=False):
141
"""Perform a diff from this to to_entry.
143
text_diff will be used for textual difference calculation.
144
This is a template method, override _diff in child classes.
146
self._read_tree_state(tree.id2path(self.file_id), tree)
148
# cannot diff from one kind to another - you must do a removal
149
# and an addif they do not match.
150
assert self.kind == to_entry.kind
151
to_entry._read_tree_state(to_tree.id2path(to_entry.file_id),
153
self._diff(text_diff, from_label, tree, to_label, to_entry, to_tree,
156
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
157
output_to, reverse=False):
158
"""Perform a diff between two entries of the same kind."""
160
def find_previous_heads(self, previous_inventories,
161
versioned_file_store,
164
"""Return the revisions and entries that directly precede this.
166
Returned as a map from revision to inventory entry.
168
This is a map containing the file revisions in all parents
169
for which the file exists, and its revision is not a parent of
170
any other. If the file is new, the set will be empty.
172
:param versioned_file_store: A store where ancestry data on this
173
file id can be queried.
174
:param transaction: The transaction that queries to the versioned
175
file store should be completed under.
176
:param entry_vf: The entry versioned file, if its already available.
178
def get_ancestors(weave, entry):
179
return set(weave.get_ancestry(entry.revision))
180
# revision:ie mapping for each ie found in previous_inventories.
182
# revision:ie mapping with one revision for each head.
184
# revision: ancestor list for each head
186
# identify candidate head revision ids.
187
for inv in previous_inventories:
188
if self.file_id in inv:
189
ie = inv[self.file_id]
190
assert ie.file_id == self.file_id
191
if ie.revision in candidates:
192
# same revision value in two different inventories:
193
# correct possible inconsistencies:
194
# * there was a bug in revision updates with 'x' bit
197
if candidates[ie.revision].executable != ie.executable:
198
candidates[ie.revision].executable = False
199
ie.executable = False
200
except AttributeError:
202
# must now be the same.
203
assert candidates[ie.revision] == ie
205
# add this revision as a candidate.
206
candidates[ie.revision] = ie
208
# common case optimisation
209
if len(candidates) == 1:
210
# if there is only one candidate revision found
211
# then we can opening the versioned file to access ancestry:
212
# there cannot be any ancestors to eliminate when there is
213
# only one revision available.
214
heads[ie.revision] = ie
217
# eliminate ancestors amongst the available candidates:
218
# heads are those that are not an ancestor of any other candidate
219
# - this provides convergence at a per-file level.
220
for ie in candidates.values():
221
# may be an ancestor of a known head:
222
already_present = 0 != len(
223
[head for head in heads
224
if ie.revision in head_ancestors[head]])
226
# an ancestor of an analyzed candidate.
228
# not an ancestor of a known head:
229
# load the versioned file for this file id if needed
231
entry_vf = versioned_file_store.get_weave_or_empty(
232
self.file_id, transaction)
233
ancestors = get_ancestors(entry_vf, ie)
234
# may knock something else out:
235
check_heads = list(heads.keys())
236
for head in check_heads:
237
if head in ancestors:
238
# this previously discovered 'head' is not
239
# really a head - its an ancestor of the newly
242
head_ancestors[ie.revision] = ancestors
243
heads[ie.revision] = ie
246
def get_tar_item(self, root, dp, now, tree):
247
"""Get a tarfile item and a file stream for its content."""
248
item = tarfile.TarInfo(pathjoin(root, dp))
249
# TODO: would be cool to actually set it to the timestamp of the
250
# revision it was last changed
252
fileobj = self._put_in_tar(item, tree)
256
"""Return true if the object this entry represents has textual data.
258
Note that textual data includes binary content.
260
Also note that all entries get weave files created for them.
261
This attribute is primarily used when upgrading from old trees that
262
did not have the weave index for all inventory entries.
266
def __init__(self, file_id, name, parent_id, text_id=None):
267
"""Create an InventoryEntry
269
The filename must be a single component, relative to the
270
parent directory; it cannot be a whole path or relative name.
272
>>> e = InventoryFile('123', 'hello.c', ROOT_ID)
277
>>> e = InventoryFile('123', 'src/hello.c', ROOT_ID)
278
Traceback (most recent call last):
279
InvalidEntryName: Invalid entry name: src/hello.c
281
assert isinstance(name, basestring), name
282
if '/' in name or '\\' in name:
283
raise InvalidEntryName(name=name)
284
self.executable = False
286
self.text_sha1 = None
287
self.text_size = None
288
self.file_id = file_id
290
self.text_id = text_id
291
self.parent_id = parent_id
292
self.symlink_target = None
294
def kind_character(self):
295
"""Return a short kind indicator useful for appending to names."""
296
raise BzrError('unknown kind %r' % self.kind)
298
known_kinds = ('file', 'directory', 'symlink')
300
def _put_in_tar(self, item, tree):
301
"""populate item for stashing in a tar, and return the content stream.
303
If no content is available, return None.
305
raise BzrError("don't know how to export {%s} of kind %r" %
306
(self.file_id, self.kind))
308
def put_on_disk(self, dest, dp, tree):
309
"""Create a representation of self on disk in the prefix dest.
311
This is a template method - implement _put_on_disk in subclasses.
313
fullpath = pathjoin(dest, dp)
314
self._put_on_disk(fullpath, tree)
315
# mutter(" export {%s} kind %s to %s", self.file_id,
316
# self.kind, fullpath)
318
def _put_on_disk(self, fullpath, tree):
319
"""Put this entry onto disk at fullpath, from tree tree."""
320
raise BzrError("don't know how to export {%s} of kind %r" % (self.file_id, self.kind))
322
def sorted_children(self):
323
return sorted(self.children.items())
326
def versionable_kind(kind):
327
return kind in ('file', 'directory', 'symlink')
329
def check(self, checker, rev_id, inv, tree):
330
"""Check this inventory entry is intact.
332
This is a template method, override _check for kind specific
335
:param checker: Check object providing context for the checks;
336
can be used to find out what parts of the repository have already
338
:param rev_id: Revision id from which this InventoryEntry was loaded.
339
Not necessarily the last-changed revision for this file.
340
:param inv: Inventory from which the entry was loaded.
341
:param tree: RevisionTree for this entry.
343
if self.parent_id is not None:
344
if not inv.has_id(self.parent_id):
345
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
346
% (self.parent_id, rev_id))
347
self._check(checker, rev_id, tree)
349
def _check(self, checker, rev_id, tree):
350
"""Check this inventory entry for kind specific errors."""
351
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
355
"""Clone this inventory entry."""
356
raise NotImplementedError
359
def describe_change(old_entry, new_entry):
360
"""Describe the change between old_entry and this.
362
This smells of being an InterInventoryEntry situation, but as its
363
the first one, we're making it a static method for now.
365
An entry with a different parent, or different name is considered
366
to be renamed. Reparenting is an internal detail.
367
Note that renaming the parent does not trigger a rename for the
370
# TODO: Perhaps return an object rather than just a string
371
if old_entry is new_entry:
372
# also the case of both being None
374
elif old_entry is None:
376
elif new_entry is None:
378
text_modified, meta_modified = new_entry.detect_changes(old_entry)
379
if text_modified or meta_modified:
383
# TODO 20060511 (mbp, rbc) factor out 'detect_rename' here.
384
if old_entry.parent_id != new_entry.parent_id:
386
elif old_entry.name != new_entry.name:
390
if renamed and not modified:
391
return InventoryEntry.RENAMED
392
if modified and not renamed:
394
if modified and renamed:
395
return InventoryEntry.MODIFIED_AND_RENAMED
399
return ("%s(%r, %r, parent_id=%r, revision=%r)"
400
% (self.__class__.__name__,
406
def snapshot(self, revision, path, previous_entries,
407
work_tree, commit_builder):
408
"""Make a snapshot of this entry which may or may not have changed.
410
This means that all its fields are populated, that it has its
411
text stored in the text store or weave.
413
# mutter('new parents of %s are %r', path, previous_entries)
414
self._read_tree_state(path, work_tree)
415
# TODO: Where should we determine whether to reuse a
416
# previous revision id or create a new revision? 20060606
417
if len(previous_entries) == 1:
418
# cannot be unchanged unless there is only one parent file rev.
419
parent_ie = previous_entries.values()[0]
420
if self._unchanged(parent_ie):
421
# mutter("found unchanged entry")
422
self.revision = parent_ie.revision
424
return self._snapshot_into_revision(revision, previous_entries,
425
work_tree, commit_builder)
427
def _snapshot_into_revision(self, revision, previous_entries, work_tree,
429
"""Record this revision unconditionally into a store.
431
The entry's last-changed revision property (`revision`) is updated to
432
that of the new revision.
434
:param revision: id of the new revision that is being recorded.
436
:returns: String description of the commit (e.g. "merged", "modified"), etc.
438
# mutter('new revision {%s} for {%s}', revision, self.file_id)
439
self.revision = revision
440
self._snapshot_text(previous_entries, work_tree, commit_builder)
442
def _snapshot_text(self, file_parents, work_tree, commit_builder):
443
"""Record the 'text' of this entry, whatever form that takes.
445
This default implementation simply adds an empty text.
447
raise NotImplementedError(self._snapshot_text)
449
def __eq__(self, other):
450
if not isinstance(other, InventoryEntry):
451
return NotImplemented
453
return ((self.file_id == other.file_id)
454
and (self.name == other.name)
455
and (other.symlink_target == self.symlink_target)
456
and (self.text_sha1 == other.text_sha1)
457
and (self.text_size == other.text_size)
458
and (self.text_id == other.text_id)
459
and (self.parent_id == other.parent_id)
460
and (self.kind == other.kind)
461
and (self.revision == other.revision)
462
and (self.executable == other.executable)
465
def __ne__(self, other):
466
return not (self == other)
469
raise ValueError('not hashable')
471
def _unchanged(self, previous_ie):
472
"""Has this entry changed relative to previous_ie.
474
This method should be overridden in child classes.
477
# different inv parent
478
if previous_ie.parent_id != self.parent_id:
481
elif previous_ie.name != self.name:
485
def _read_tree_state(self, path, work_tree):
486
"""Populate fields in the inventory entry from the given tree.
488
Note that this should be modified to be a noop on virtual trees
489
as all entries created there are prepopulated.
491
# TODO: Rather than running this manually, we should check the
492
# working sha1 and other expensive properties when they're
493
# first requested, or preload them if they're already known
494
pass # nothing to do by default
496
def _forget_tree_state(self):
500
class InventoryDirectory(InventoryEntry):
501
"""A directory in an inventory."""
503
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
504
'text_id', 'parent_id', 'children', 'executable',
505
'revision', 'symlink_target']
507
def _check(self, checker, rev_id, tree):
508
"""See InventoryEntry._check"""
509
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
510
raise BzrCheckError('directory {%s} has text in revision {%s}'
511
% (self.file_id, rev_id))
514
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
515
other.revision = self.revision
516
# note that children are *not* copied; they're pulled across when
520
def __init__(self, file_id, name, parent_id):
521
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
523
self.kind = 'directory'
525
def kind_character(self):
526
"""See InventoryEntry.kind_character."""
529
def _put_in_tar(self, item, tree):
530
"""See InventoryEntry._put_in_tar."""
531
item.type = tarfile.DIRTYPE
538
def _put_on_disk(self, fullpath, tree):
539
"""See InventoryEntry._put_on_disk."""
542
def _snapshot_text(self, file_parents, work_tree, commit_builder):
543
"""See InventoryEntry._snapshot_text."""
544
commit_builder.modified_directory(self.file_id, file_parents)
547
class InventoryFile(InventoryEntry):
548
"""A file in an inventory."""
550
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
551
'text_id', 'parent_id', 'children', 'executable',
552
'revision', 'symlink_target']
554
def _check(self, checker, tree_revision_id, tree):
555
"""See InventoryEntry._check"""
556
t = (self.file_id, self.revision)
557
if t in checker.checked_texts:
558
prev_sha = checker.checked_texts[t]
559
if prev_sha != self.text_sha1:
560
raise BzrCheckError('mismatched sha1 on {%s} in {%s}' %
561
(self.file_id, tree_revision_id))
563
checker.repeated_text_cnt += 1
566
if self.file_id not in checker.checked_weaves:
567
mutter('check weave {%s}', self.file_id)
568
w = tree.get_weave(self.file_id)
569
# Not passing a progress bar, because it creates a new
570
# progress, which overwrites the current progress,
571
# and doesn't look nice
573
checker.checked_weaves[self.file_id] = True
575
w = tree.get_weave(self.file_id)
577
mutter('check version {%s} of {%s}', tree_revision_id, self.file_id)
578
checker.checked_text_cnt += 1
579
# We can't check the length, because Weave doesn't store that
580
# information, and the whole point of looking at the weave's
581
# sha1sum is that we don't have to extract the text.
582
if self.text_sha1 != w.get_sha1(self.revision):
583
raise BzrCheckError('text {%s} version {%s} wrong sha1'
584
% (self.file_id, self.revision))
585
checker.checked_texts[t] = self.text_sha1
588
other = InventoryFile(self.file_id, self.name, self.parent_id)
589
other.executable = self.executable
590
other.text_id = self.text_id
591
other.text_sha1 = self.text_sha1
592
other.text_size = self.text_size
593
other.revision = self.revision
596
def detect_changes(self, old_entry):
597
"""See InventoryEntry.detect_changes."""
598
assert self.text_sha1 is not None
599
assert old_entry.text_sha1 is not None
600
text_modified = (self.text_sha1 != old_entry.text_sha1)
601
meta_modified = (self.executable != old_entry.executable)
602
return text_modified, meta_modified
604
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
605
output_to, reverse=False):
606
"""See InventoryEntry._diff."""
608
from_text = tree.get_file(self.file_id).readlines()
610
to_text = to_tree.get_file(to_entry.file_id).readlines()
614
text_diff(from_label, from_text,
615
to_label, to_text, output_to)
617
text_diff(to_label, to_text,
618
from_label, from_text, output_to)
621
label_pair = (to_label, from_label)
623
label_pair = (from_label, to_label)
624
print >> output_to, "Binary files %s and %s differ" % label_pair
627
"""See InventoryEntry.has_text."""
630
def __init__(self, file_id, name, parent_id):
631
super(InventoryFile, self).__init__(file_id, name, parent_id)
634
def kind_character(self):
635
"""See InventoryEntry.kind_character."""
638
def _put_in_tar(self, item, tree):
639
"""See InventoryEntry._put_in_tar."""
640
item.type = tarfile.REGTYPE
641
fileobj = tree.get_file(self.file_id)
642
item.size = self.text_size
643
if tree.is_executable(self.file_id):
649
def _put_on_disk(self, fullpath, tree):
650
"""See InventoryEntry._put_on_disk."""
651
pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
652
if tree.is_executable(self.file_id):
653
os.chmod(fullpath, 0755)
655
def _read_tree_state(self, path, work_tree):
656
"""See InventoryEntry._read_tree_state."""
657
self.text_sha1 = work_tree.get_file_sha1(self.file_id, path=path)
658
# FIXME: 20050930 probe for the text size when getting sha1
659
# in _read_tree_state
660
self.executable = work_tree.is_executable(self.file_id, path=path)
663
return ("%s(%r, %r, parent_id=%r, sha1=%r, len=%s)"
664
% (self.__class__.__name__,
671
def _forget_tree_state(self):
672
self.text_sha1 = None
674
def _snapshot_text(self, file_parents, work_tree, commit_builder):
675
"""See InventoryEntry._snapshot_text."""
676
def get_content_byte_lines():
677
return work_tree.get_file(self.file_id).readlines()
678
self.text_sha1, self.text_size = commit_builder.modified_file_text(
679
self.file_id, file_parents, get_content_byte_lines, self.text_sha1, self.text_size)
681
def _unchanged(self, previous_ie):
682
"""See InventoryEntry._unchanged."""
683
compatible = super(InventoryFile, self)._unchanged(previous_ie)
684
if self.text_sha1 != previous_ie.text_sha1:
687
# FIXME: 20050930 probe for the text size when getting sha1
688
# in _read_tree_state
689
self.text_size = previous_ie.text_size
690
if self.executable != previous_ie.executable:
695
class InventoryLink(InventoryEntry):
696
"""A file in an inventory."""
698
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
699
'text_id', 'parent_id', 'children', 'executable',
700
'revision', 'symlink_target']
702
def _check(self, checker, rev_id, tree):
703
"""See InventoryEntry._check"""
704
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
705
raise BzrCheckError('symlink {%s} has text in revision {%s}'
706
% (self.file_id, rev_id))
707
if self.symlink_target is None:
708
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
709
% (self.file_id, rev_id))
712
other = InventoryLink(self.file_id, self.name, self.parent_id)
713
other.symlink_target = self.symlink_target
714
other.revision = self.revision
717
def detect_changes(self, old_entry):
718
"""See InventoryEntry.detect_changes."""
719
# FIXME: which _modified field should we use ? RBC 20051003
720
text_modified = (self.symlink_target != old_entry.symlink_target)
722
mutter(" symlink target changed")
723
meta_modified = False
724
return text_modified, meta_modified
726
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
727
output_to, reverse=False):
728
"""See InventoryEntry._diff."""
729
from_text = self.symlink_target
730
if to_entry is not None:
731
to_text = to_entry.symlink_target
736
print >>output_to, '=== target changed %r => %r' % (from_text, to_text)
739
print >>output_to, '=== target was %r' % self.symlink_target
741
print >>output_to, '=== target is %r' % self.symlink_target
743
def __init__(self, file_id, name, parent_id):
744
super(InventoryLink, self).__init__(file_id, name, parent_id)
745
self.kind = 'symlink'
747
def kind_character(self):
748
"""See InventoryEntry.kind_character."""
751
def _put_in_tar(self, item, tree):
752
"""See InventoryEntry._put_in_tar."""
753
item.type = tarfile.SYMTYPE
757
item.linkname = self.symlink_target
760
def _put_on_disk(self, fullpath, tree):
761
"""See InventoryEntry._put_on_disk."""
763
os.symlink(self.symlink_target, fullpath)
765
raise BzrError("Failed to create symlink %r -> %r, error: %s" % (fullpath, self.symlink_target, e))
767
def _read_tree_state(self, path, work_tree):
768
"""See InventoryEntry._read_tree_state."""
769
self.symlink_target = work_tree.get_symlink_target(self.file_id)
771
def _forget_tree_state(self):
772
self.symlink_target = None
774
def _unchanged(self, previous_ie):
775
"""See InventoryEntry._unchanged."""
776
compatible = super(InventoryLink, self)._unchanged(previous_ie)
777
if self.symlink_target != previous_ie.symlink_target:
781
def _snapshot_text(self, file_parents, work_tree, commit_builder):
782
"""See InventoryEntry._snapshot_text."""
783
commit_builder.modified_link(
784
self.file_id, file_parents, self.symlink_target)
787
class Inventory(object):
788
"""Inventory of versioned files in a tree.
790
This describes which file_id is present at each point in the tree,
791
and possibly the SHA-1 or other information about the file.
792
Entries can be looked up either by path or by file_id.
794
The inventory represents a typical unix file tree, with
795
directories containing files and subdirectories. We never store
796
the full path to a file, because renaming a directory implicitly
797
moves all of its contents. This class internally maintains a
798
lookup tree that allows the children under a directory to be
801
InventoryEntry objects must not be modified after they are
802
inserted, other than through the Inventory API.
804
>>> inv = Inventory()
805
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
806
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
807
>>> inv['123-123'].name
810
May be treated as an iterator or set to look up file ids:
812
>>> bool(inv.path2id('hello.c'))
817
May also look up by name:
819
>>> [x[0] for x in inv.iter_entries()]
821
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
822
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
823
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT-12345678-12345678', sha1=None, len=None)
825
def __init__(self, root_id=ROOT_ID, revision_id=None):
826
"""Create or read an inventory.
828
If a working directory is specified, the inventory is read
829
from there. If the file is specified, read from that. If not,
830
the inventory is created empty.
832
The inventory is created with a default root directory, with
835
# We are letting Branch.create() create a unique inventory
836
# root id. Rather than generating a random one here.
838
# root_id = bzrlib.branch.gen_file_id('TREE_ROOT')
839
self.root = InventoryDirectory(root_id, '', None)
840
# FIXME: this isn't ever used, changing it to self.revision may break
841
# things. TODO make everything use self.revision_id
842
self.revision_id = revision_id
843
self._byid = {self.root.file_id: self.root}
846
# TODO: jam 20051218 Should copy also copy the revision_id?
847
entries = self.iter_entries()
848
other = Inventory(entries.next()[1].file_id)
849
# copy recursively so we know directories will be added before
850
# their children. There are more efficient ways than this...
851
for path, entry in entries():
852
other.add(entry.copy())
856
return iter(self._byid)
859
"""Returns number of entries."""
860
return len(self._byid)
862
def iter_entries(self, from_dir=None):
863
"""Return (path, entry) pairs, in order by name."""
868
elif isinstance(from_dir, basestring):
869
from_dir = self._byid[from_dir]
871
# unrolling the recursive called changed the time from
872
# 440ms/663ms (inline/total) to 116ms/116ms
873
children = from_dir.children.items()
875
children = collections.deque(children)
876
stack = [(u'', children)]
878
from_dir_relpath, children = stack[-1]
881
name, ie = children.popleft()
883
# we know that from_dir_relpath never ends in a slash
884
# and 'f' doesn't begin with one, we can do a string op, rather
885
# than the checks of pathjoin(), though this means that all paths
887
path = from_dir_relpath + '/' + name
891
if ie.kind != 'directory':
894
# But do this child first
895
new_children = ie.children.items()
897
new_children = collections.deque(new_children)
898
stack.append((path, new_children))
899
# Break out of inner loop, so that we start outer loop with child
902
# if we finished all children, pop it off the stack
905
def iter_entries_by_dir(self, from_dir=None):
906
"""Iterate over the entries in a directory first order.
908
This returns all entries for a directory before returning
909
the entries for children of a directory. This is not
910
lexicographically sorted order, and is a hybrid between
911
depth-first and breadth-first.
913
:return: This yields (path, entry) pairs
915
# TODO? Perhaps this should return the from_dir so that the root is
916
# yielded? or maybe an option?
921
elif isinstance(from_dir, basestring):
922
from_dir = self._byid[from_dir]
924
stack = [(u'', from_dir)]
926
cur_relpath, cur_dir = stack.pop()
929
for child_name, child_ie in sorted(cur_dir.children.iteritems()):
931
child_relpath = cur_relpath + child_name
933
yield child_relpath, child_ie
935
if child_ie.kind == 'directory':
936
child_dirs.append((child_relpath+'/', child_ie))
937
stack.extend(reversed(child_dirs))
940
"""Return list of (path, ie) for all entries except the root.
942
This may be faster than iter_entries.
945
def descend(dir_ie, dir_path):
946
kids = dir_ie.children.items()
948
for name, ie in kids:
949
child_path = pathjoin(dir_path, name)
950
accum.append((child_path, ie))
951
if ie.kind == 'directory':
952
descend(ie, child_path)
954
descend(self.root, u'')
957
def directories(self):
958
"""Return (path, entry) pairs for all directories, including the root.
961
def descend(parent_ie, parent_path):
962
accum.append((parent_path, parent_ie))
964
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
967
for name, child_ie in kids:
968
child_path = pathjoin(parent_path, name)
969
descend(child_ie, child_path)
970
descend(self.root, u'')
973
def __contains__(self, file_id):
974
"""True if this entry contains a file with given id.
976
>>> inv = Inventory()
977
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
978
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
984
return file_id in self._byid
986
def __getitem__(self, file_id):
987
"""Return the entry for given file_id.
989
>>> inv = Inventory()
990
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
991
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
992
>>> inv['123123'].name
996
return self._byid[file_id]
999
raise BzrError("can't look up file_id None")
1001
raise BzrError("file_id {%s} not in inventory" % file_id)
1003
def get_file_kind(self, file_id):
1004
return self._byid[file_id].kind
1006
def get_child(self, parent_id, filename):
1007
return self[parent_id].children.get(filename)
1009
def add(self, entry):
1010
"""Add entry to inventory.
1012
To add a file to a branch ready to be committed, use Branch.add,
1015
Returns the new entry object.
1017
if entry.file_id in self._byid:
1018
raise BzrError("inventory already contains entry with id {%s}" % entry.file_id)
1020
if entry.parent_id == ROOT_ID or entry.parent_id is None:
1021
entry.parent_id = self.root.file_id
1024
parent = self._byid[entry.parent_id]
1026
raise BzrError("parent_id {%s} not in inventory" % entry.parent_id)
1028
if entry.name in parent.children:
1029
raise BzrError("%s is already versioned" %
1030
pathjoin(self.id2path(parent.file_id), entry.name))
1032
self._byid[entry.file_id] = entry
1033
parent.children[entry.name] = entry
1036
def add_path(self, relpath, kind, file_id=None, parent_id=None):
1037
"""Add entry from a path.
1039
The immediate parent must already be versioned.
1041
Returns the new entry object."""
1043
parts = osutils.splitpath(relpath)
1047
file_id = bzrlib.workingtree.gen_root_id()
1048
self.root = InventoryDirectory(file_id, '', None)
1049
self._byid = {self.root.file_id: self.root}
1052
parent_path = parts[:-1]
1053
parent_id = self.path2id(parent_path)
1054
if parent_id is None:
1055
raise NotVersionedError(path=parent_path)
1056
ie = make_entry(kind, parts[-1], parent_id, file_id)
1059
def __delitem__(self, file_id):
1060
"""Remove entry by id.
1062
>>> inv = Inventory()
1063
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1064
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
1073
assert ie.parent_id is None or \
1074
self[ie.parent_id].children[ie.name] == ie
1076
del self._byid[file_id]
1077
if ie.parent_id is not None:
1078
del self[ie.parent_id].children[ie.name]
1080
def __eq__(self, other):
1081
"""Compare two sets by comparing their contents.
1083
>>> i1 = Inventory()
1084
>>> i2 = Inventory()
1087
>>> i1.add(InventoryFile('123', 'foo', ROOT_ID))
1088
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
1091
>>> i2.add(InventoryFile('123', 'foo', ROOT_ID))
1092
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
1096
if not isinstance(other, Inventory):
1097
return NotImplemented
1099
return self._byid == other._byid
1101
def __ne__(self, other):
1102
return not self.__eq__(other)
1105
raise ValueError('not hashable')
1107
def _iter_file_id_parents(self, file_id):
1108
"""Yield the parents of file_id up to the root."""
1109
while file_id is not None:
1111
ie = self._byid[file_id]
1113
raise BzrError("file_id {%s} not found in inventory" % file_id)
1115
file_id = ie.parent_id
1117
def get_idpath(self, file_id):
1118
"""Return a list of file_ids for the path to an entry.
1120
The list contains one element for each directory followed by
1121
the id of the file itself. So the length of the returned list
1122
is equal to the depth of the file in the tree, counting the
1123
root directory as depth 1.
1126
for parent in self._iter_file_id_parents(file_id):
1127
p.insert(0, parent.file_id)
1130
def id2path(self, file_id):
1131
"""Return as a string the path to file_id.
1134
>>> e = i.add(InventoryDirectory('src-id', 'src', ROOT_ID))
1135
>>> e = i.add(InventoryFile('foo-id', 'foo.c', parent_id='src-id'))
1136
>>> print i.id2path('foo-id')
1139
# get all names, skipping root
1140
return '/'.join(reversed(
1141
[parent.name for parent in
1142
self._iter_file_id_parents(file_id)][:-1]))
1144
def path2id(self, name):
1145
"""Walk down through directories to return entry of last component.
1147
names may be either a list of path components, or a single
1148
string, in which case it is automatically split.
1150
This returns the entry of the last component in the path,
1151
which may be either a file or a directory.
1153
Returns None IFF the path is not found.
1155
if isinstance(name, types.StringTypes):
1156
name = splitpath(name)
1158
# mutter("lookup path %r" % name)
1163
cie = parent.children[f]
1164
assert cie.name == f
1165
assert cie.parent_id == parent.file_id
1168
# or raise an error?
1171
return parent.file_id
1173
def has_filename(self, names):
1174
return bool(self.path2id(names))
1176
def has_id(self, file_id):
1177
return self._byid.has_key(file_id)
1179
def rename(self, file_id, new_parent_id, new_name):
1180
"""Move a file within the inventory.
1182
This can change either the name, or the parent, or both.
1184
This does not move the working file."""
1185
if not is_valid_name(new_name):
1186
raise BzrError("not an acceptable filename: %r" % new_name)
1188
new_parent = self._byid[new_parent_id]
1189
if new_name in new_parent.children:
1190
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
1192
new_parent_idpath = self.get_idpath(new_parent_id)
1193
if file_id in new_parent_idpath:
1194
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
1195
% (self.id2path(file_id), self.id2path(new_parent_id)))
1197
file_ie = self._byid[file_id]
1198
old_parent = self._byid[file_ie.parent_id]
1200
# TODO: Don't leave things messed up if this fails
1202
del old_parent.children[file_ie.name]
1203
new_parent.children[new_name] = file_ie
1205
file_ie.name = new_name
1206
file_ie.parent_id = new_parent_id
1208
def is_root(self, file_id):
1209
return self.root is not None and file_id == self.root.file_id
1212
def make_entry(kind, name, parent_id, file_id=None):
1213
"""Create an inventory entry.
1215
:param kind: the type of inventory entry to create.
1216
:param name: the basename of the entry.
1217
:param parent_id: the parent_id of the entry.
1218
:param file_id: the file_id to use. if None, one will be created.
1221
file_id = bzrlib.workingtree.gen_file_id(name)
1223
norm_name, can_access = osutils.normalized_filename(name)
1224
if norm_name != name:
1228
# TODO: jam 20060701 This would probably be more useful
1229
# if the error was raised with the full path
1230
raise errors.InvalidNormalization(name)
1232
if kind == 'directory':
1233
return InventoryDirectory(file_id, name, parent_id)
1234
elif kind == 'file':
1235
return InventoryFile(file_id, name, parent_id)
1236
elif kind == 'symlink':
1237
return InventoryLink(file_id, name, parent_id)
1239
raise BzrError("unknown kind %r" % kind)
1244
def is_valid_name(name):
1246
if _NAME_RE is None:
1247
_NAME_RE = re.compile(r'^[^/\\]+$')
1249
return bool(_NAME_RE.match(name))
added
removed
bzrlib/inventory.py
