1
# (C) 2005 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.
38
from bzrlib.osutils import (pumpfile, quotefn, splitpath, joinpath,
39
pathjoin, sha_strings)
40
from bzrlib.errors import (NotVersionedError, InvalidEntryName,
41
BzrError, BzrCheckError, BinaryFile)
42
from bzrlib.trace import mutter
45
class InventoryEntry(object):
46
"""Description of a versioned file.
48
An InventoryEntry has the following fields, which are also
49
present in the XML inventory-entry element:
54
(within the parent directory)
57
file_id of the parent directory, or ROOT_ID
60
the revision_id in which this variation of this file was
64
Indicates that this file should be executable on systems
68
sha-1 of the text of the file
71
size in bytes of the text of the file
73
(reading a version 4 tree created a text_id field.)
78
>>> i.add(InventoryDirectory('123', 'src', ROOT_ID))
79
InventoryDirectory('123', 'src', parent_id='TREE_ROOT')
80
>>> i.add(InventoryFile('2323', 'hello.c', parent_id='123'))
81
InventoryFile('2323', 'hello.c', parent_id='123')
82
>>> shouldbe = {0: 'src', 1: pathjoin('src','hello.c')}
83
>>> for ix, j in enumerate(i.iter_entries()):
84
... print (j[0] == shouldbe[ix], j[1])
86
(True, InventoryDirectory('123', 'src', parent_id='TREE_ROOT'))
87
(True, InventoryFile('2323', 'hello.c', parent_id='123'))
88
>>> i.add(InventoryFile('2323', 'bye.c', '123'))
89
Traceback (most recent call last):
91
BzrError: inventory already contains entry with id {2323}
92
>>> i.add(InventoryFile('2324', 'bye.c', '123'))
93
InventoryFile('2324', 'bye.c', parent_id='123')
94
>>> i.add(InventoryDirectory('2325', 'wibble', '123'))
95
InventoryDirectory('2325', 'wibble', parent_id='123')
96
>>> i.path2id('src/wibble')
100
>>> i.add(InventoryFile('2326', 'wibble.c', '2325'))
101
InventoryFile('2326', 'wibble.c', parent_id='2325')
103
InventoryFile('2326', 'wibble.c', parent_id='2325')
104
>>> for path, entry in i.iter_entries():
106
... assert i.path2id(path)
113
>>> i.id2path('2326')
114
'src/wibble/wibble.c'
117
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
118
'text_id', 'parent_id', 'children', 'executable',
121
def _add_text_to_weave(self, new_lines, parents, weave_store, transaction):
122
versionedfile = weave_store.get_weave_or_empty(self.file_id,
124
versionedfile.add_lines(self.revision, parents, new_lines)
125
versionedfile.clear_cache()
127
def detect_changes(self, old_entry):
128
"""Return a (text_modified, meta_modified) from this to old_entry.
130
_read_tree_state must have been called on self and old_entry prior to
131
calling detect_changes.
135
def diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
136
output_to, reverse=False):
137
"""Perform a diff from this to to_entry.
139
text_diff will be used for textual difference calculation.
140
This is a template method, override _diff in child classes.
142
self._read_tree_state(tree.id2path(self.file_id), tree)
144
# cannot diff from one kind to another - you must do a removal
145
# and an addif they do not match.
146
assert self.kind == to_entry.kind
147
to_entry._read_tree_state(to_tree.id2path(to_entry.file_id),
149
self._diff(text_diff, from_label, tree, to_label, to_entry, to_tree,
152
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
153
output_to, reverse=False):
154
"""Perform a diff between two entries of the same kind."""
156
def find_previous_heads(self, previous_inventories,
157
versioned_file_store,
160
"""Return the revisions and entries that directly preceed this.
162
Returned as a map from revision to inventory entry.
164
This is a map containing the file revisions in all parents
165
for which the file exists, and its revision is not a parent of
166
any other. If the file is new, the set will be empty.
168
:param versioned_file_store: A store where ancestry data on this
169
file id can be queried.
170
:param transaction: The transaction that queries to the versioned
171
file store should be completed under.
172
:param entry_vf: The entry versioned file, if its already available.
174
def get_ancestors(weave, entry):
175
return set(weave.get_ancestry(entry.revision))
176
# revision:ie mapping for each ie found in previous_inventories.
178
# revision:ie mapping with one revision for each head.
180
# revision: ancestor list for each head
182
# identify candidate head revision ids.
183
for inv in previous_inventories:
184
if self.file_id in inv:
185
ie = inv[self.file_id]
186
assert ie.file_id == self.file_id
187
if ie.revision in candidates:
188
# same revision value in two different inventories:
189
# correct possible inconsistencies:
190
# * there was a bug in revision updates with 'x' bit
193
if candidates[ie.revision].executable != ie.executable:
194
candidates[ie.revision].executable = False
195
ie.executable = False
196
except AttributeError:
198
# must now be the same.
199
assert candidates[ie.revision] == ie
201
# add this revision as a candidate.
202
candidates[ie.revision] = ie
204
# common case optimisation
205
if len(candidates) == 1:
206
# if there is only one candidate revision found
207
# then we can opening the versioned file to access ancestry:
208
# there cannot be any ancestors to eliminate when there is
209
# only one revision available.
210
heads[ie.revision] = ie
213
# eliminate ancestors amongst the available candidates:
214
# heads are those that are not an ancestor of any other candidate
215
# - this provides convergence at a per-file level.
216
for ie in candidates.values():
217
# may be an ancestor of a known head:
218
already_present = 0 != len(
219
[head for head in heads
220
if ie.revision in head_ancestors[head]])
222
# an ancestor of an analyzed candidate.
224
# not an ancestor of a known head:
225
# load the versioned file for this file id if needed
227
entry_vf = versioned_file_store.get_weave_or_empty(
228
self.file_id, transaction)
229
ancestors = get_ancestors(entry_vf, ie)
230
# may knock something else out:
231
check_heads = list(heads.keys())
232
for head in check_heads:
233
if head in ancestors:
234
# this previously discovered 'head' is not
235
# really a head - its an ancestor of the newly
238
head_ancestors[ie.revision] = ancestors
239
heads[ie.revision] = ie
242
def get_tar_item(self, root, dp, now, tree):
243
"""Get a tarfile item and a file stream for its content."""
244
item = tarfile.TarInfo(pathjoin(root, dp))
245
# TODO: would be cool to actually set it to the timestamp of the
246
# revision it was last changed
248
fileobj = self._put_in_tar(item, tree)
252
"""Return true if the object this entry represents has textual data.
254
Note that textual data includes binary content.
256
Also note that all entries get weave files created for them.
257
This attribute is primarily used when upgrading from old trees that
258
did not have the weave index for all inventory entries.
262
def __init__(self, file_id, name, parent_id, text_id=None):
263
"""Create an InventoryEntry
265
The filename must be a single component, relative to the
266
parent directory; it cannot be a whole path or relative name.
268
>>> e = InventoryFile('123', 'hello.c', ROOT_ID)
273
>>> e = InventoryFile('123', 'src/hello.c', ROOT_ID)
274
Traceback (most recent call last):
275
InvalidEntryName: Invalid entry name: src/hello.c
277
assert isinstance(name, basestring), name
278
if '/' in name or '\\' in name:
279
raise InvalidEntryName(name=name)
280
self.executable = False
282
self.text_sha1 = None
283
self.text_size = None
284
self.file_id = file_id
286
self.text_id = text_id
287
self.parent_id = parent_id
288
self.symlink_target = None
290
def kind_character(self):
291
"""Return a short kind indicator useful for appending to names."""
292
raise BzrError('unknown kind %r' % self.kind)
294
known_kinds = ('file', 'directory', 'symlink', 'root_directory')
296
def _put_in_tar(self, item, tree):
297
"""populate item for stashing in a tar, and return the content stream.
299
If no content is available, return None.
301
raise BzrError("don't know how to export {%s} of kind %r" %
302
(self.file_id, self.kind))
304
def put_on_disk(self, dest, dp, tree):
305
"""Create a representation of self on disk in the prefix dest.
307
This is a template method - implement _put_on_disk in subclasses.
309
fullpath = pathjoin(dest, dp)
310
self._put_on_disk(fullpath, tree)
311
mutter(" export {%s} kind %s to %s", self.file_id,
314
def _put_on_disk(self, fullpath, tree):
315
"""Put this entry onto disk at fullpath, from tree tree."""
316
raise BzrError("don't know how to export {%s} of kind %r" % (self.file_id, self.kind))
318
def sorted_children(self):
319
l = self.children.items()
324
def versionable_kind(kind):
325
return kind in ('file', 'directory', 'symlink')
327
def check(self, checker, rev_id, inv, tree):
328
"""Check this inventory entry is intact.
330
This is a template method, override _check for kind specific
333
:param checker: Check object providing context for the checks;
334
can be used to find out what parts of the repository have already
336
:param rev_id: Revision id from which this InventoryEntry was loaded.
337
Not necessarily the last-changed revision for this file.
338
:param inv: Inventory from which the entry was loaded.
339
:param tree: RevisionTree for this entry.
341
if self.parent_id != None:
342
if not inv.has_id(self.parent_id):
343
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
344
% (self.parent_id, rev_id))
345
self._check(checker, rev_id, tree)
347
def _check(self, checker, rev_id, tree):
348
"""Check this inventory entry for kind specific errors."""
349
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
354
"""Clone this inventory entry."""
355
raise NotImplementedError
357
def _get_snapshot_change(self, previous_entries):
358
if len(previous_entries) > 1:
360
elif len(previous_entries) == 0:
363
return 'modified/renamed/reparented'
366
return ("%s(%r, %r, parent_id=%r)"
367
% (self.__class__.__name__,
372
def snapshot(self, revision, path, previous_entries,
373
work_tree, weave_store, transaction):
374
"""Make a snapshot of this entry which may or may not have changed.
376
This means that all its fields are populated, that it has its
377
text stored in the text store or weave.
379
mutter('new parents of %s are %r', path, previous_entries)
380
self._read_tree_state(path, work_tree)
381
if len(previous_entries) == 1:
382
# cannot be unchanged unless there is only one parent file rev.
383
parent_ie = previous_entries.values()[0]
384
if self._unchanged(parent_ie):
385
mutter("found unchanged entry")
386
self.revision = parent_ie.revision
388
return self.snapshot_revision(revision, previous_entries,
389
work_tree, weave_store, transaction)
391
def snapshot_revision(self, revision, previous_entries, work_tree,
392
weave_store, transaction):
393
"""Record this revision unconditionally."""
394
mutter('new revision for {%s}', self.file_id)
395
self.revision = revision
396
change = self._get_snapshot_change(previous_entries)
397
self._snapshot_text(previous_entries, work_tree, weave_store,
401
def _snapshot_text(self, file_parents, work_tree, weave_store, transaction):
402
"""Record the 'text' of this entry, whatever form that takes.
404
This default implementation simply adds an empty text.
406
mutter('storing file {%s} in revision {%s}',
407
self.file_id, self.revision)
408
self._add_text_to_weave([], file_parents.keys(), weave_store, transaction)
410
def __eq__(self, other):
411
if not isinstance(other, InventoryEntry):
412
return NotImplemented
414
return ((self.file_id == other.file_id)
415
and (self.name == other.name)
416
and (other.symlink_target == self.symlink_target)
417
and (self.text_sha1 == other.text_sha1)
418
and (self.text_size == other.text_size)
419
and (self.text_id == other.text_id)
420
and (self.parent_id == other.parent_id)
421
and (self.kind == other.kind)
422
and (self.revision == other.revision)
423
and (self.executable == other.executable)
426
def __ne__(self, other):
427
return not (self == other)
430
raise ValueError('not hashable')
432
def _unchanged(self, previous_ie):
433
"""Has this entry changed relative to previous_ie.
435
This method should be overriden in child classes.
438
# different inv parent
439
if previous_ie.parent_id != self.parent_id:
442
elif previous_ie.name != self.name:
446
def _read_tree_state(self, path, work_tree):
447
"""Populate fields in the inventory entry from the given tree.
449
Note that this should be modified to be a noop on virtual trees
450
as all entries created there are prepopulated.
452
# TODO: Rather than running this manually, we should check the
453
# working sha1 and other expensive properties when they're
454
# first requested, or preload them if they're already known
455
pass # nothing to do by default
457
def _forget_tree_state(self):
461
class RootEntry(InventoryEntry):
463
def _check(self, checker, rev_id, tree):
464
"""See InventoryEntry._check"""
466
def __init__(self, file_id):
467
self.file_id = file_id
469
self.kind = 'root_directory'
470
self.parent_id = None
473
def __eq__(self, other):
474
if not isinstance(other, RootEntry):
475
return NotImplemented
477
return (self.file_id == other.file_id) \
478
and (self.children == other.children)
481
class InventoryDirectory(InventoryEntry):
482
"""A directory in an inventory."""
484
def _check(self, checker, rev_id, tree):
485
"""See InventoryEntry._check"""
486
if self.text_sha1 != None or self.text_size != None or self.text_id != None:
487
raise BzrCheckError('directory {%s} has text in revision {%s}'
488
% (self.file_id, rev_id))
491
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
492
other.revision = self.revision
493
# note that children are *not* copied; they're pulled across when
497
def __init__(self, file_id, name, parent_id):
498
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
500
self.kind = 'directory'
502
def kind_character(self):
503
"""See InventoryEntry.kind_character."""
506
def _put_in_tar(self, item, tree):
507
"""See InventoryEntry._put_in_tar."""
508
item.type = tarfile.DIRTYPE
515
def _put_on_disk(self, fullpath, tree):
516
"""See InventoryEntry._put_on_disk."""
520
class InventoryFile(InventoryEntry):
521
"""A file in an inventory."""
523
def _check(self, checker, tree_revision_id, tree):
524
"""See InventoryEntry._check"""
525
t = (self.file_id, self.revision)
526
if t in checker.checked_texts:
527
prev_sha = checker.checked_texts[t]
528
if prev_sha != self.text_sha1:
529
raise BzrCheckError('mismatched sha1 on {%s} in {%s}' %
530
(self.file_id, tree_revision_id))
532
checker.repeated_text_cnt += 1
535
if self.file_id not in checker.checked_weaves:
536
mutter('check weave {%s}', self.file_id)
537
w = tree.get_weave(self.file_id)
538
# Not passing a progress bar, because it creates a new
539
# progress, which overwrites the current progress,
540
# and doesn't look nice
542
checker.checked_weaves[self.file_id] = True
544
w = tree.get_weave(self.file_id)
546
mutter('check version {%s} of {%s}', tree_revision_id, self.file_id)
547
checker.checked_text_cnt += 1
548
# We can't check the length, because Weave doesn't store that
549
# information, and the whole point of looking at the weave's
550
# sha1sum is that we don't have to extract the text.
551
if self.text_sha1 != w.get_sha1(self.revision):
552
raise BzrCheckError('text {%s} version {%s} wrong sha1'
553
% (self.file_id, self.revision))
554
checker.checked_texts[t] = self.text_sha1
557
other = InventoryFile(self.file_id, self.name, self.parent_id)
558
other.executable = self.executable
559
other.text_id = self.text_id
560
other.text_sha1 = self.text_sha1
561
other.text_size = self.text_size
562
other.revision = self.revision
565
def detect_changes(self, old_entry):
566
"""See InventoryEntry.detect_changes."""
567
assert self.text_sha1 != None
568
assert old_entry.text_sha1 != None
569
text_modified = (self.text_sha1 != old_entry.text_sha1)
570
meta_modified = (self.executable != old_entry.executable)
571
return text_modified, meta_modified
573
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
574
output_to, reverse=False):
575
"""See InventoryEntry._diff."""
577
from_text = tree.get_file(self.file_id).readlines()
579
to_text = to_tree.get_file(to_entry.file_id).readlines()
583
text_diff(from_label, from_text,
584
to_label, to_text, output_to)
586
text_diff(to_label, to_text,
587
from_label, from_text, output_to)
590
label_pair = (to_label, from_label)
592
label_pair = (from_label, to_label)
593
print >> output_to, "Binary files %s and %s differ" % label_pair
596
"""See InventoryEntry.has_text."""
599
def __init__(self, file_id, name, parent_id):
600
super(InventoryFile, self).__init__(file_id, name, parent_id)
603
def kind_character(self):
604
"""See InventoryEntry.kind_character."""
607
def _put_in_tar(self, item, tree):
608
"""See InventoryEntry._put_in_tar."""
609
item.type = tarfile.REGTYPE
610
fileobj = tree.get_file(self.file_id)
611
item.size = self.text_size
612
if tree.is_executable(self.file_id):
618
def _put_on_disk(self, fullpath, tree):
619
"""See InventoryEntry._put_on_disk."""
620
pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
621
if tree.is_executable(self.file_id):
622
os.chmod(fullpath, 0755)
624
def _read_tree_state(self, path, work_tree):
625
"""See InventoryEntry._read_tree_state."""
626
self.text_sha1 = work_tree.get_file_sha1(self.file_id)
627
self.executable = work_tree.is_executable(self.file_id)
629
def _forget_tree_state(self):
630
self.text_sha1 = None
631
self.executable = None
633
def _snapshot_text(self, file_parents, work_tree, weave_store, transaction):
634
"""See InventoryEntry._snapshot_text."""
635
mutter('storing file {%s} in revision {%s}',
636
self.file_id, self.revision)
637
# special case to avoid diffing on renames or
639
if (len(file_parents) == 1
640
and self.text_sha1 == file_parents.values()[0].text_sha1
641
and self.text_size == file_parents.values()[0].text_size):
642
previous_ie = file_parents.values()[0]
643
versionedfile = weave_store.get_weave(self.file_id, transaction)
644
versionedfile.clone_text(self.revision, previous_ie.revision, file_parents.keys())
646
new_lines = work_tree.get_file(self.file_id).readlines()
647
self._add_text_to_weave(new_lines, file_parents.keys(), weave_store,
649
self.text_sha1 = sha_strings(new_lines)
650
self.text_size = sum(map(len, new_lines))
653
def _unchanged(self, previous_ie):
654
"""See InventoryEntry._unchanged."""
655
compatible = super(InventoryFile, self)._unchanged(previous_ie)
656
if self.text_sha1 != previous_ie.text_sha1:
659
# FIXME: 20050930 probe for the text size when getting sha1
660
# in _read_tree_state
661
self.text_size = previous_ie.text_size
662
if self.executable != previous_ie.executable:
667
class InventoryLink(InventoryEntry):
668
"""A file in an inventory."""
670
__slots__ = ['symlink_target']
672
def _check(self, checker, rev_id, tree):
673
"""See InventoryEntry._check"""
674
if self.text_sha1 != None or self.text_size != None or self.text_id != None:
675
raise BzrCheckError('symlink {%s} has text in revision {%s}'
676
% (self.file_id, rev_id))
677
if self.symlink_target == None:
678
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
679
% (self.file_id, rev_id))
682
other = InventoryLink(self.file_id, self.name, self.parent_id)
683
other.symlink_target = self.symlink_target
684
other.revision = self.revision
687
def detect_changes(self, old_entry):
688
"""See InventoryEntry.detect_changes."""
689
# FIXME: which _modified field should we use ? RBC 20051003
690
text_modified = (self.symlink_target != old_entry.symlink_target)
692
mutter(" symlink target changed")
693
meta_modified = False
694
return text_modified, meta_modified
696
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
697
output_to, reverse=False):
698
"""See InventoryEntry._diff."""
699
from_text = self.symlink_target
700
if to_entry is not None:
701
to_text = to_entry.symlink_target
706
print >>output_to, '=== target changed %r => %r' % (from_text, to_text)
709
print >>output_to, '=== target was %r' % self.symlink_target
711
print >>output_to, '=== target is %r' % self.symlink_target
713
def __init__(self, file_id, name, parent_id):
714
super(InventoryLink, self).__init__(file_id, name, parent_id)
715
self.kind = 'symlink'
717
def kind_character(self):
718
"""See InventoryEntry.kind_character."""
721
def _put_in_tar(self, item, tree):
722
"""See InventoryEntry._put_in_tar."""
723
item.type = tarfile.SYMTYPE
727
item.linkname = self.symlink_target
730
def _put_on_disk(self, fullpath, tree):
731
"""See InventoryEntry._put_on_disk."""
733
os.symlink(self.symlink_target, fullpath)
735
raise BzrError("Failed to create symlink %r -> %r, error: %s" % (fullpath, self.symlink_target, e))
737
def _read_tree_state(self, path, work_tree):
738
"""See InventoryEntry._read_tree_state."""
739
self.symlink_target = work_tree.get_symlink_target(self.file_id)
741
def _forget_tree_state(self):
742
self.symlink_target = None
744
def _unchanged(self, previous_ie):
745
"""See InventoryEntry._unchanged."""
746
compatible = super(InventoryLink, self)._unchanged(previous_ie)
747
if self.symlink_target != previous_ie.symlink_target:
752
class Inventory(object):
753
"""Inventory of versioned files in a tree.
755
This describes which file_id is present at each point in the tree,
756
and possibly the SHA-1 or other information about the file.
757
Entries can be looked up either by path or by file_id.
759
The inventory represents a typical unix file tree, with
760
directories containing files and subdirectories. We never store
761
the full path to a file, because renaming a directory implicitly
762
moves all of its contents. This class internally maintains a
763
lookup tree that allows the children under a directory to be
766
InventoryEntry objects must not be modified after they are
767
inserted, other than through the Inventory API.
769
>>> inv = Inventory()
770
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
771
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT')
772
>>> inv['123-123'].name
775
May be treated as an iterator or set to look up file ids:
777
>>> bool(inv.path2id('hello.c'))
782
May also look up by name:
784
>>> [x[0] for x in inv.iter_entries()]
786
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
787
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
788
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT-12345678-12345678')
790
def __init__(self, root_id=ROOT_ID, revision_id=None):
791
"""Create or read an inventory.
793
If a working directory is specified, the inventory is read
794
from there. If the file is specified, read from that. If not,
795
the inventory is created empty.
797
The inventory is created with a default root directory, with
800
# We are letting Branch.create() create a unique inventory
801
# root id. Rather than generating a random one here.
803
# root_id = bzrlib.branch.gen_file_id('TREE_ROOT')
804
self.root = RootEntry(root_id)
805
self.revision_id = revision_id
806
self._byid = {self.root.file_id: self.root}
810
# TODO: jam 20051218 Should copy also copy the revision_id?
811
other = Inventory(self.root.file_id)
812
# copy recursively so we know directories will be added before
813
# their children. There are more efficient ways than this...
814
for path, entry in self.iter_entries():
815
if entry == self.root:
817
other.add(entry.copy())
822
return iter(self._byid)
826
"""Returns number of entries."""
827
return len(self._byid)
830
def iter_entries(self, from_dir=None):
831
"""Return (path, entry) pairs, in order by name."""
835
elif isinstance(from_dir, basestring):
836
from_dir = self._byid[from_dir]
838
kids = from_dir.children.items()
840
for name, ie in kids:
842
if ie.kind == 'directory':
843
for cn, cie in self.iter_entries(from_dir=ie.file_id):
844
yield pathjoin(name, cn), cie
848
"""Return list of (path, ie) for all entries except the root.
850
This may be faster than iter_entries.
853
def descend(dir_ie, dir_path):
854
kids = dir_ie.children.items()
856
for name, ie in kids:
857
child_path = pathjoin(dir_path, name)
858
accum.append((child_path, ie))
859
if ie.kind == 'directory':
860
descend(ie, child_path)
862
descend(self.root, u'')
866
def directories(self):
867
"""Return (path, entry) pairs for all directories, including the root.
870
def descend(parent_ie, parent_path):
871
accum.append((parent_path, parent_ie))
873
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
876
for name, child_ie in kids:
877
child_path = pathjoin(parent_path, name)
878
descend(child_ie, child_path)
879
descend(self.root, u'')
884
def __contains__(self, file_id):
885
"""True if this entry contains a file with given id.
887
>>> inv = Inventory()
888
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
889
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT')
895
return file_id in self._byid
898
def __getitem__(self, file_id):
899
"""Return the entry for given file_id.
901
>>> inv = Inventory()
902
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
903
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT')
904
>>> inv['123123'].name
908
return self._byid[file_id]
911
raise BzrError("can't look up file_id None")
913
raise BzrError("file_id {%s} not in inventory" % file_id)
916
def get_file_kind(self, file_id):
917
return self._byid[file_id].kind
919
def get_child(self, parent_id, filename):
920
return self[parent_id].children.get(filename)
923
def add(self, entry):
924
"""Add entry to inventory.
926
To add a file to a branch ready to be committed, use Branch.add,
929
Returns the new entry object.
931
if entry.file_id in self._byid:
932
raise BzrError("inventory already contains entry with id {%s}" % entry.file_id)
934
if entry.parent_id == ROOT_ID or entry.parent_id is None:
935
entry.parent_id = self.root.file_id
938
parent = self._byid[entry.parent_id]
940
raise BzrError("parent_id {%s} not in inventory" % entry.parent_id)
942
if parent.children.has_key(entry.name):
943
raise BzrError("%s is already versioned" %
944
pathjoin(self.id2path(parent.file_id), entry.name))
946
self._byid[entry.file_id] = entry
947
parent.children[entry.name] = entry
951
def add_path(self, relpath, kind, file_id=None):
952
"""Add entry from a path.
954
The immediate parent must already be versioned.
956
Returns the new entry object."""
957
from bzrlib.workingtree import gen_file_id
959
parts = bzrlib.osutils.splitpath(relpath)
962
file_id = gen_file_id(relpath)
965
self.root = RootEntry(file_id)
966
self._byid = {self.root.file_id: self.root}
969
parent_path = parts[:-1]
970
parent_id = self.path2id(parent_path)
971
if parent_id == None:
972
raise NotVersionedError(path=parent_path)
973
if kind == 'directory':
974
ie = InventoryDirectory(file_id, parts[-1], parent_id)
976
ie = InventoryFile(file_id, parts[-1], parent_id)
977
elif kind == 'symlink':
978
ie = InventoryLink(file_id, parts[-1], parent_id)
980
raise BzrError("unknown kind %r" % kind)
984
def __delitem__(self, file_id):
985
"""Remove entry by id.
987
>>> inv = Inventory()
988
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
989
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT')
998
assert ie.parent_id is None or \
999
self[ie.parent_id].children[ie.name] == ie
1001
del self._byid[file_id]
1002
if ie.parent_id is not None:
1003
del self[ie.parent_id].children[ie.name]
1006
def __eq__(self, other):
1007
"""Compare two sets by comparing their contents.
1009
>>> i1 = Inventory()
1010
>>> i2 = Inventory()
1013
>>> i1.add(InventoryFile('123', 'foo', ROOT_ID))
1014
InventoryFile('123', 'foo', parent_id='TREE_ROOT')
1017
>>> i2.add(InventoryFile('123', 'foo', ROOT_ID))
1018
InventoryFile('123', 'foo', parent_id='TREE_ROOT')
1022
if not isinstance(other, Inventory):
1023
return NotImplemented
1025
if len(self._byid) != len(other._byid):
1026
# shortcut: obviously not the same
1029
return self._byid == other._byid
1032
def __ne__(self, other):
1033
return not self.__eq__(other)
1037
raise ValueError('not hashable')
1039
def _iter_file_id_parents(self, file_id):
1040
"""Yield the parents of file_id up to the root."""
1041
while file_id != None:
1043
ie = self._byid[file_id]
1045
raise BzrError("file_id {%s} not found in inventory" % file_id)
1047
file_id = ie.parent_id
1049
def get_idpath(self, file_id):
1050
"""Return a list of file_ids for the path to an entry.
1052
The list contains one element for each directory followed by
1053
the id of the file itself. So the length of the returned list
1054
is equal to the depth of the file in the tree, counting the
1055
root directory as depth 1.
1058
for parent in self._iter_file_id_parents(file_id):
1059
p.insert(0, parent.file_id)
1062
def id2path(self, file_id):
1063
"""Return as a string the path to file_id.
1066
>>> e = i.add(InventoryDirectory('src-id', 'src', ROOT_ID))
1067
>>> e = i.add(InventoryFile('foo-id', 'foo.c', parent_id='src-id'))
1068
>>> print i.id2path('foo-id')
1071
# get all names, skipping root
1072
return '/'.join(reversed(
1073
[parent.name for parent in
1074
self._iter_file_id_parents(file_id)][:-1]))
1076
def path2id(self, name):
1077
"""Walk down through directories to return entry of last component.
1079
names may be either a list of path components, or a single
1080
string, in which case it is automatically split.
1082
This returns the entry of the last component in the path,
1083
which may be either a file or a directory.
1085
Returns None iff the path is not found.
1087
if isinstance(name, types.StringTypes):
1088
name = splitpath(name)
1090
mutter("lookup path %r" % name)
1095
cie = parent.children[f]
1096
assert cie.name == f
1097
assert cie.parent_id == parent.file_id
1100
# or raise an error?
1103
return parent.file_id
1106
def has_filename(self, names):
1107
return bool(self.path2id(names))
1110
def has_id(self, file_id):
1111
return self._byid.has_key(file_id)
1114
def rename(self, file_id, new_parent_id, new_name):
1115
"""Move a file within the inventory.
1117
This can change either the name, or the parent, or both.
1119
This does not move the working file."""
1120
if not is_valid_name(new_name):
1121
raise BzrError("not an acceptable filename: %r" % new_name)
1123
new_parent = self._byid[new_parent_id]
1124
if new_name in new_parent.children:
1125
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
1127
new_parent_idpath = self.get_idpath(new_parent_id)
1128
if file_id in new_parent_idpath:
1129
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
1130
% (self.id2path(file_id), self.id2path(new_parent_id)))
1132
file_ie = self._byid[file_id]
1133
old_parent = self._byid[file_ie.parent_id]
1135
# TODO: Don't leave things messed up if this fails
1137
del old_parent.children[file_ie.name]
1138
new_parent.children[new_name] = file_ie
1140
file_ie.name = new_name
1141
file_ie.parent_id = new_parent_id
1148
def is_valid_name(name):
1150
if _NAME_RE == None:
1151
_NAME_RE = re.compile(r'^[^/\\]+$')
1153
return bool(_NAME_RE.match(name))
added
removed
bzrlib/inventory.py
