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.trace import mutter
41
from bzrlib.errors import (NotVersionedError, InvalidEntryName,
42
BzrError, BzrCheckError)
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)
126
def detect_changes(self, old_entry):
127
"""Return a (text_modified, meta_modified) from this to old_entry.
129
_read_tree_state must have been called on self and old_entry prior to
130
calling detect_changes.
134
def diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
135
output_to, reverse=False):
136
"""Perform a diff from this to to_entry.
138
text_diff will be used for textual difference calculation.
139
This is a template method, override _diff in child classes.
141
self._read_tree_state(tree.id2path(self.file_id), tree)
143
# cannot diff from one kind to another - you must do a removal
144
# and an addif they do not match.
145
assert self.kind == to_entry.kind
146
to_entry._read_tree_state(to_tree.id2path(to_entry.file_id),
148
self._diff(text_diff, from_label, tree, to_label, to_entry, to_tree,
151
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
152
output_to, reverse=False):
153
"""Perform a diff between two entries of the same kind."""
155
def find_previous_heads(self, previous_inventories,
156
versioned_file_store,
159
"""Return the revisions and entries that directly preceed this.
161
Returned as a map from revision to inventory entry.
163
This is a map containing the file revisions in all parents
164
for which the file exists, and its revision is not a parent of
165
any other. If the file is new, the set will be empty.
167
:param versioned_file_store: A store where ancestry data on this
168
file id can be queried.
169
:param transaction: The transaction that queries to the versioned
170
file store should be completed under.
171
:param entry_vf: The entry versioned file, if its already available.
173
def get_ancestors(weave, entry):
174
return set(weave.get_ancestry(entry.revision))
175
# revision:ie mapping for each ie found in previous_inventories.
177
# revision:ie mapping with one revision for each head.
179
# revision: ancestor list for each head
181
# identify candidate head revision ids.
182
for inv in previous_inventories:
183
if self.file_id in inv:
184
ie = inv[self.file_id]
185
assert ie.file_id == self.file_id
186
if ie.revision in candidates:
187
# same revision value in two different inventories:
188
# correct possible inconsistencies:
189
# * there was a bug in revision updates with 'x' bit
192
if candidates[ie.revision].executable != ie.executable:
193
candidates[ie.revision].executable = False
194
ie.executable = False
195
except AttributeError:
197
# must now be the same.
198
assert candidates[ie.revision] == ie
200
# add this revision as a candidate.
201
candidates[ie.revision] = ie
203
# common case optimisation
204
if len(candidates) == 1:
205
# if there is only one candidate revision found
206
# then we can opening the versioned file to access ancestry:
207
# there cannot be any ancestors to eliminate when there is
208
# only one revision available.
209
heads[ie.revision] = ie
212
# eliminate ancestors amongst the available candidates:
213
# heads are those that are not an ancestor of any other candidate
214
# - this provides convergence at a per-file level.
215
for ie in candidates.values():
216
# may be an ancestor of a known head:
217
already_present = 0 != len(
218
[head for head in heads
219
if ie.revision in head_ancestors[head]])
221
# an ancestor of an analyzed candidate.
223
# not an ancestor of a known head:
224
# load the versioned file for this file id if needed
226
entry_vf = versioned_file_store.get_weave_or_empty(
227
self.file_id, transaction)
228
ancestors = get_ancestors(entry_vf, ie)
229
# may knock something else out:
230
check_heads = list(heads.keys())
231
for head in check_heads:
232
if head in ancestors:
233
# this previously discovered 'head' is not
234
# really a head - its an ancestor of the newly
237
head_ancestors[ie.revision] = ancestors
238
heads[ie.revision] = ie
241
def get_tar_item(self, root, dp, now, tree):
242
"""Get a tarfile item and a file stream for its content."""
243
item = tarfile.TarInfo(pathjoin(root, dp))
244
# TODO: would be cool to actually set it to the timestamp of the
245
# revision it was last changed
247
fileobj = self._put_in_tar(item, tree)
251
"""Return true if the object this entry represents has textual data.
253
Note that textual data includes binary content.
255
Also note that all entries get weave files created for them.
256
This attribute is primarily used when upgrading from old trees that
257
did not have the weave index for all inventory entries.
261
def __init__(self, file_id, name, parent_id, text_id=None):
262
"""Create an InventoryEntry
264
The filename must be a single component, relative to the
265
parent directory; it cannot be a whole path or relative name.
267
>>> e = InventoryFile('123', 'hello.c', ROOT_ID)
272
>>> e = InventoryFile('123', 'src/hello.c', ROOT_ID)
273
Traceback (most recent call last):
274
InvalidEntryName: Invalid entry name: src/hello.c
276
assert isinstance(name, basestring), name
277
if '/' in name or '\\' in name:
278
raise InvalidEntryName(name=name)
279
self.executable = False
281
self.text_sha1 = None
282
self.text_size = None
283
self.file_id = file_id
285
self.text_id = text_id
286
self.parent_id = parent_id
287
self.symlink_target = None
289
def kind_character(self):
290
"""Return a short kind indicator useful for appending to names."""
291
raise BzrError('unknown kind %r' % self.kind)
293
known_kinds = ('file', 'directory', 'symlink', 'root_directory')
295
def _put_in_tar(self, item, tree):
296
"""populate item for stashing in a tar, and return the content stream.
298
If no content is available, return None.
300
raise BzrError("don't know how to export {%s} of kind %r" %
301
(self.file_id, self.kind))
303
def put_on_disk(self, dest, dp, tree):
304
"""Create a representation of self on disk in the prefix dest.
306
This is a template method - implement _put_on_disk in subclasses.
308
fullpath = pathjoin(dest, dp)
309
self._put_on_disk(fullpath, tree)
310
mutter(" export {%s} kind %s to %s", self.file_id,
313
def _put_on_disk(self, fullpath, tree):
314
"""Put this entry onto disk at fullpath, from tree tree."""
315
raise BzrError("don't know how to export {%s} of kind %r" % (self.file_id, self.kind))
317
def sorted_children(self):
318
l = self.children.items()
323
def versionable_kind(kind):
324
return kind in ('file', 'directory', 'symlink')
326
def check(self, checker, rev_id, inv, tree):
327
"""Check this inventory entry is intact.
329
This is a template method, override _check for kind specific
332
if self.parent_id != None:
333
if not inv.has_id(self.parent_id):
334
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
335
% (self.parent_id, rev_id))
336
self._check(checker, rev_id, tree)
338
def _check(self, checker, rev_id, tree):
339
"""Check this inventory entry for kind specific errors."""
340
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
345
"""Clone this inventory entry."""
346
raise NotImplementedError
348
def _get_snapshot_change(self, previous_entries):
349
if len(previous_entries) > 1:
351
elif len(previous_entries) == 0:
354
return 'modified/renamed/reparented'
357
return ("%s(%r, %r, parent_id=%r)"
358
% (self.__class__.__name__,
363
def snapshot(self, revision, path, previous_entries,
364
work_tree, weave_store, transaction):
365
"""Make a snapshot of this entry which may or may not have changed.
367
This means that all its fields are populated, that it has its
368
text stored in the text store or weave.
370
mutter('new parents of %s are %r', path, previous_entries)
371
self._read_tree_state(path, work_tree)
372
if len(previous_entries) == 1:
373
# cannot be unchanged unless there is only one parent file rev.
374
parent_ie = previous_entries.values()[0]
375
if self._unchanged(parent_ie):
376
mutter("found unchanged entry")
377
self.revision = parent_ie.revision
379
return self.snapshot_revision(revision, previous_entries,
380
work_tree, weave_store, transaction)
382
def snapshot_revision(self, revision, previous_entries, work_tree,
383
weave_store, transaction):
384
"""Record this revision unconditionally."""
385
mutter('new revision for {%s}', self.file_id)
386
self.revision = revision
387
change = self._get_snapshot_change(previous_entries)
388
self._snapshot_text(previous_entries, work_tree, weave_store,
392
def _snapshot_text(self, file_parents, work_tree, weave_store, transaction):
393
"""Record the 'text' of this entry, whatever form that takes.
395
This default implementation simply adds an empty text.
397
mutter('storing file {%s} in revision {%s}',
398
self.file_id, self.revision)
399
self._add_text_to_weave([], file_parents.keys(), weave_store, transaction)
401
def __eq__(self, other):
402
if not isinstance(other, InventoryEntry):
403
return NotImplemented
405
return ((self.file_id == other.file_id)
406
and (self.name == other.name)
407
and (other.symlink_target == self.symlink_target)
408
and (self.text_sha1 == other.text_sha1)
409
and (self.text_size == other.text_size)
410
and (self.text_id == other.text_id)
411
and (self.parent_id == other.parent_id)
412
and (self.kind == other.kind)
413
and (self.revision == other.revision)
414
and (self.executable == other.executable)
417
def __ne__(self, other):
418
return not (self == other)
421
raise ValueError('not hashable')
423
def _unchanged(self, previous_ie):
424
"""Has this entry changed relative to previous_ie.
426
This method should be overriden in child classes.
429
# different inv parent
430
if previous_ie.parent_id != self.parent_id:
433
elif previous_ie.name != self.name:
437
def _read_tree_state(self, path, work_tree):
438
"""Populate fields in the inventory entry from the given tree.
440
Note that this should be modified to be a noop on virtual trees
441
as all entries created there are prepopulated.
443
# TODO: Rather than running this manually, we should check the
444
# working sha1 and other expensive properties when they're
445
# first requested, or preload them if they're already known
446
pass # nothing to do by default
448
def _forget_tree_state(self):
452
class RootEntry(InventoryEntry):
454
def _check(self, checker, rev_id, tree):
455
"""See InventoryEntry._check"""
457
def __init__(self, file_id):
458
self.file_id = file_id
460
self.kind = 'root_directory'
461
self.parent_id = None
464
def __eq__(self, other):
465
if not isinstance(other, RootEntry):
466
return NotImplemented
468
return (self.file_id == other.file_id) \
469
and (self.children == other.children)
472
class InventoryDirectory(InventoryEntry):
473
"""A directory in an inventory."""
475
def _check(self, checker, rev_id, tree):
476
"""See InventoryEntry._check"""
477
if self.text_sha1 != None or self.text_size != None or self.text_id != None:
478
raise BzrCheckError('directory {%s} has text in revision {%s}'
479
% (self.file_id, rev_id))
482
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
483
other.revision = self.revision
484
# note that children are *not* copied; they're pulled across when
488
def __init__(self, file_id, name, parent_id):
489
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
491
self.kind = 'directory'
493
def kind_character(self):
494
"""See InventoryEntry.kind_character."""
497
def _put_in_tar(self, item, tree):
498
"""See InventoryEntry._put_in_tar."""
499
item.type = tarfile.DIRTYPE
506
def _put_on_disk(self, fullpath, tree):
507
"""See InventoryEntry._put_on_disk."""
511
class InventoryFile(InventoryEntry):
512
"""A file in an inventory."""
514
def _check(self, checker, rev_id, tree):
515
"""See InventoryEntry._check"""
516
revision = self.revision
517
t = (self.file_id, revision)
518
if t in checker.checked_texts:
519
prev_sha = checker.checked_texts[t]
520
if prev_sha != self.text_sha1:
521
raise BzrCheckError('mismatched sha1 on {%s} in {%s}' %
522
(self.file_id, rev_id))
524
checker.repeated_text_cnt += 1
527
if self.file_id not in checker.checked_weaves:
528
mutter('check weave {%s}', self.file_id)
529
w = tree.get_weave(self.file_id)
530
# Not passing a progress bar, because it creates a new
531
# progress, which overwrites the current progress,
532
# and doesn't look nice
534
checker.checked_weaves[self.file_id] = True
536
w = tree.get_weave(self.file_id)
538
mutter('check version {%s} of {%s}', rev_id, self.file_id)
539
checker.checked_text_cnt += 1
540
# We can't check the length, because Weave doesn't store that
541
# information, and the whole point of looking at the weave's
542
# sha1sum is that we don't have to extract the text.
543
if self.text_sha1 != w.get_sha1(self.revision):
544
raise BzrCheckError('text {%s} version {%s} wrong sha1'
545
% (self.file_id, self.revision))
546
checker.checked_texts[t] = self.text_sha1
549
other = InventoryFile(self.file_id, self.name, self.parent_id)
550
other.executable = self.executable
551
other.text_id = self.text_id
552
other.text_sha1 = self.text_sha1
553
other.text_size = self.text_size
554
other.revision = self.revision
557
def detect_changes(self, old_entry):
558
"""See InventoryEntry.detect_changes."""
559
assert self.text_sha1 != None
560
assert old_entry.text_sha1 != None
561
text_modified = (self.text_sha1 != old_entry.text_sha1)
562
meta_modified = (self.executable != old_entry.executable)
563
return text_modified, meta_modified
565
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
566
output_to, reverse=False):
567
"""See InventoryEntry._diff."""
568
from_text = tree.get_file(self.file_id).readlines()
570
to_text = to_tree.get_file(to_entry.file_id).readlines()
574
text_diff(from_label, from_text,
575
to_label, to_text, output_to)
577
text_diff(to_label, to_text,
578
from_label, from_text, output_to)
581
"""See InventoryEntry.has_text."""
584
def __init__(self, file_id, name, parent_id):
585
super(InventoryFile, self).__init__(file_id, name, parent_id)
588
def kind_character(self):
589
"""See InventoryEntry.kind_character."""
592
def _put_in_tar(self, item, tree):
593
"""See InventoryEntry._put_in_tar."""
594
item.type = tarfile.REGTYPE
595
fileobj = tree.get_file(self.file_id)
596
item.size = self.text_size
597
if tree.is_executable(self.file_id):
603
def _put_on_disk(self, fullpath, tree):
604
"""See InventoryEntry._put_on_disk."""
605
pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
606
if tree.is_executable(self.file_id):
607
os.chmod(fullpath, 0755)
609
def _read_tree_state(self, path, work_tree):
610
"""See InventoryEntry._read_tree_state."""
611
self.text_sha1 = work_tree.get_file_sha1(self.file_id)
612
self.executable = work_tree.is_executable(self.file_id)
614
def _forget_tree_state(self):
615
self.text_sha1 = None
616
self.executable = None
618
def _snapshot_text(self, file_parents, work_tree, weave_store, transaction):
619
"""See InventoryEntry._snapshot_text."""
620
mutter('storing file {%s} in revision {%s}',
621
self.file_id, self.revision)
622
# special case to avoid diffing on renames or
624
if (len(file_parents) == 1
625
and self.text_sha1 == file_parents.values()[0].text_sha1
626
and self.text_size == file_parents.values()[0].text_size):
627
previous_ie = file_parents.values()[0]
628
versionedfile = weave_store.get_weave(self.file_id, transaction)
629
versionedfile.clone_text(self.revision, previous_ie.revision, file_parents.keys())
631
new_lines = work_tree.get_file(self.file_id).readlines()
632
self._add_text_to_weave(new_lines, file_parents.keys(), weave_store,
634
self.text_sha1 = sha_strings(new_lines)
635
self.text_size = sum(map(len, new_lines))
638
def _unchanged(self, previous_ie):
639
"""See InventoryEntry._unchanged."""
640
compatible = super(InventoryFile, self)._unchanged(previous_ie)
641
if self.text_sha1 != previous_ie.text_sha1:
644
# FIXME: 20050930 probe for the text size when getting sha1
645
# in _read_tree_state
646
self.text_size = previous_ie.text_size
647
if self.executable != previous_ie.executable:
652
class InventoryLink(InventoryEntry):
653
"""A file in an inventory."""
655
__slots__ = ['symlink_target']
657
def _check(self, checker, rev_id, tree):
658
"""See InventoryEntry._check"""
659
if self.text_sha1 != None or self.text_size != None or self.text_id != None:
660
raise BzrCheckError('symlink {%s} has text in revision {%s}'
661
% (self.file_id, rev_id))
662
if self.symlink_target == None:
663
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
664
% (self.file_id, rev_id))
667
other = InventoryLink(self.file_id, self.name, self.parent_id)
668
other.symlink_target = self.symlink_target
669
other.revision = self.revision
672
def detect_changes(self, old_entry):
673
"""See InventoryEntry.detect_changes."""
674
# FIXME: which _modified field should we use ? RBC 20051003
675
text_modified = (self.symlink_target != old_entry.symlink_target)
677
mutter(" symlink target changed")
678
meta_modified = False
679
return text_modified, meta_modified
681
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
682
output_to, reverse=False):
683
"""See InventoryEntry._diff."""
684
from_text = self.symlink_target
685
if to_entry is not None:
686
to_text = to_entry.symlink_target
691
print >>output_to, '=== target changed %r => %r' % (from_text, to_text)
694
print >>output_to, '=== target was %r' % self.symlink_target
696
print >>output_to, '=== target is %r' % self.symlink_target
698
def __init__(self, file_id, name, parent_id):
699
super(InventoryLink, self).__init__(file_id, name, parent_id)
700
self.kind = 'symlink'
702
def kind_character(self):
703
"""See InventoryEntry.kind_character."""
706
def _put_in_tar(self, item, tree):
707
"""See InventoryEntry._put_in_tar."""
708
item.type = tarfile.SYMTYPE
712
item.linkname = self.symlink_target
715
def _put_on_disk(self, fullpath, tree):
716
"""See InventoryEntry._put_on_disk."""
718
os.symlink(self.symlink_target, fullpath)
720
raise BzrError("Failed to create symlink %r -> %r, error: %s" % (fullpath, self.symlink_target, e))
722
def _read_tree_state(self, path, work_tree):
723
"""See InventoryEntry._read_tree_state."""
724
self.symlink_target = work_tree.get_symlink_target(self.file_id)
726
def _forget_tree_state(self):
727
self.symlink_target = None
729
def _unchanged(self, previous_ie):
730
"""See InventoryEntry._unchanged."""
731
compatible = super(InventoryLink, self)._unchanged(previous_ie)
732
if self.symlink_target != previous_ie.symlink_target:
737
class Inventory(object):
738
"""Inventory of versioned files in a tree.
740
This describes which file_id is present at each point in the tree,
741
and possibly the SHA-1 or other information about the file.
742
Entries can be looked up either by path or by file_id.
744
The inventory represents a typical unix file tree, with
745
directories containing files and subdirectories. We never store
746
the full path to a file, because renaming a directory implicitly
747
moves all of its contents. This class internally maintains a
748
lookup tree that allows the children under a directory to be
751
InventoryEntry objects must not be modified after they are
752
inserted, other than through the Inventory API.
754
>>> inv = Inventory()
755
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
756
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT')
757
>>> inv['123-123'].name
760
May be treated as an iterator or set to look up file ids:
762
>>> bool(inv.path2id('hello.c'))
767
May also look up by name:
769
>>> [x[0] for x in inv.iter_entries()]
771
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
772
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
773
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT-12345678-12345678')
775
def __init__(self, root_id=ROOT_ID):
776
"""Create or read an inventory.
778
If a working directory is specified, the inventory is read
779
from there. If the file is specified, read from that. If not,
780
the inventory is created empty.
782
The inventory is created with a default root directory, with
785
# We are letting Branch.create() create a unique inventory
786
# root id. Rather than generating a random one here.
788
# root_id = bzrlib.branch.gen_file_id('TREE_ROOT')
789
self.root = RootEntry(root_id)
790
self._byid = {self.root.file_id: self.root}
794
other = Inventory(self.root.file_id)
795
# copy recursively so we know directories will be added before
796
# their children. There are more efficient ways than this...
797
for path, entry in self.iter_entries():
798
if entry == self.root:
800
other.add(entry.copy())
805
return iter(self._byid)
809
"""Returns number of entries."""
810
return len(self._byid)
813
def iter_entries(self, from_dir=None):
814
"""Return (path, entry) pairs, in order by name."""
818
elif isinstance(from_dir, basestring):
819
from_dir = self._byid[from_dir]
821
kids = from_dir.children.items()
823
for name, ie in kids:
825
if ie.kind == 'directory':
826
for cn, cie in self.iter_entries(from_dir=ie.file_id):
827
yield pathjoin(name, cn), cie
831
"""Return list of (path, ie) for all entries except the root.
833
This may be faster than iter_entries.
836
def descend(dir_ie, dir_path):
837
kids = dir_ie.children.items()
839
for name, ie in kids:
840
child_path = pathjoin(dir_path, name)
841
accum.append((child_path, ie))
842
if ie.kind == 'directory':
843
descend(ie, child_path)
845
descend(self.root, u'')
849
def directories(self):
850
"""Return (path, entry) pairs for all directories, including the root.
853
def descend(parent_ie, parent_path):
854
accum.append((parent_path, parent_ie))
856
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
859
for name, child_ie in kids:
860
child_path = pathjoin(parent_path, name)
861
descend(child_ie, child_path)
862
descend(self.root, u'')
867
def __contains__(self, file_id):
868
"""True if this entry contains a file with given id.
870
>>> inv = Inventory()
871
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
872
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT')
878
return file_id in self._byid
881
def __getitem__(self, file_id):
882
"""Return the entry for given file_id.
884
>>> inv = Inventory()
885
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
886
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT')
887
>>> inv['123123'].name
891
return self._byid[file_id]
894
raise BzrError("can't look up file_id None")
896
raise BzrError("file_id {%s} not in inventory" % file_id)
899
def get_file_kind(self, file_id):
900
return self._byid[file_id].kind
902
def get_child(self, parent_id, filename):
903
return self[parent_id].children.get(filename)
906
def add(self, entry):
907
"""Add entry to inventory.
909
To add a file to a branch ready to be committed, use Branch.add,
912
Returns the new entry object.
914
if entry.file_id in self._byid:
915
raise BzrError("inventory already contains entry with id {%s}" % entry.file_id)
917
if entry.parent_id == ROOT_ID or entry.parent_id is None:
918
entry.parent_id = self.root.file_id
921
parent = self._byid[entry.parent_id]
923
raise BzrError("parent_id {%s} not in inventory" % entry.parent_id)
925
if parent.children.has_key(entry.name):
926
raise BzrError("%s is already versioned" %
927
pathjoin(self.id2path(parent.file_id), entry.name))
929
self._byid[entry.file_id] = entry
930
parent.children[entry.name] = entry
934
def add_path(self, relpath, kind, file_id=None):
935
"""Add entry from a path.
937
The immediate parent must already be versioned.
939
Returns the new entry object."""
940
from bzrlib.workingtree import gen_file_id
942
parts = bzrlib.osutils.splitpath(relpath)
945
file_id = gen_file_id(relpath)
948
self.root = RootEntry(file_id)
949
self._byid = {self.root.file_id: self.root}
952
parent_path = parts[:-1]
953
parent_id = self.path2id(parent_path)
954
if parent_id == None:
955
raise NotVersionedError(path=parent_path)
956
if kind == 'directory':
957
ie = InventoryDirectory(file_id, parts[-1], parent_id)
959
ie = InventoryFile(file_id, parts[-1], parent_id)
960
elif kind == 'symlink':
961
ie = InventoryLink(file_id, parts[-1], parent_id)
963
raise BzrError("unknown kind %r" % kind)
967
def __delitem__(self, file_id):
968
"""Remove entry by id.
970
>>> inv = Inventory()
971
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
972
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT')
981
assert ie.parent_id is None or \
982
self[ie.parent_id].children[ie.name] == ie
984
del self._byid[file_id]
985
if ie.parent_id is not None:
986
del self[ie.parent_id].children[ie.name]
989
def __eq__(self, other):
990
"""Compare two sets by comparing their contents.
996
>>> i1.add(InventoryFile('123', 'foo', ROOT_ID))
997
InventoryFile('123', 'foo', parent_id='TREE_ROOT')
1000
>>> i2.add(InventoryFile('123', 'foo', ROOT_ID))
1001
InventoryFile('123', 'foo', parent_id='TREE_ROOT')
1005
if not isinstance(other, Inventory):
1006
return NotImplemented
1008
if len(self._byid) != len(other._byid):
1009
# shortcut: obviously not the same
1012
return self._byid == other._byid
1015
def __ne__(self, other):
1016
return not self.__eq__(other)
1020
raise ValueError('not hashable')
1022
def _iter_file_id_parents(self, file_id):
1023
"""Yield the parents of file_id up to the root."""
1024
while file_id != None:
1026
ie = self._byid[file_id]
1028
raise BzrError("file_id {%s} not found in inventory" % file_id)
1030
file_id = ie.parent_id
1032
def get_idpath(self, file_id):
1033
"""Return a list of file_ids for the path to an entry.
1035
The list contains one element for each directory followed by
1036
the id of the file itself. So the length of the returned list
1037
is equal to the depth of the file in the tree, counting the
1038
root directory as depth 1.
1041
for parent in self._iter_file_id_parents(file_id):
1042
p.insert(0, parent.file_id)
1045
def id2path(self, file_id):
1046
"""Return as a string the path to file_id.
1049
>>> e = i.add(InventoryDirectory('src-id', 'src', ROOT_ID))
1050
>>> e = i.add(InventoryFile('foo-id', 'foo.c', parent_id='src-id'))
1051
>>> print i.id2path('foo-id')
1054
# get all names, skipping root
1055
return '/'.join(reversed(
1056
[parent.name for parent in
1057
self._iter_file_id_parents(file_id)][:-1]))
1059
def path2id(self, name):
1060
"""Walk down through directories to return entry of last component.
1062
names may be either a list of path components, or a single
1063
string, in which case it is automatically split.
1065
This returns the entry of the last component in the path,
1066
which may be either a file or a directory.
1068
Returns None iff the path is not found.
1070
if isinstance(name, types.StringTypes):
1071
name = splitpath(name)
1073
mutter("lookup path %r" % name)
1078
cie = parent.children[f]
1079
assert cie.name == f
1080
assert cie.parent_id == parent.file_id
1083
# or raise an error?
1086
return parent.file_id
1089
def has_filename(self, names):
1090
return bool(self.path2id(names))
1093
def has_id(self, file_id):
1094
return self._byid.has_key(file_id)
1097
def rename(self, file_id, new_parent_id, new_name):
1098
"""Move a file within the inventory.
1100
This can change either the name, or the parent, or both.
1102
This does not move the working file."""
1103
if not is_valid_name(new_name):
1104
raise BzrError("not an acceptable filename: %r" % new_name)
1106
new_parent = self._byid[new_parent_id]
1107
if new_name in new_parent.children:
1108
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
1110
new_parent_idpath = self.get_idpath(new_parent_id)
1111
if file_id in new_parent_idpath:
1112
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
1113
% (self.id2path(file_id), self.id2path(new_parent_id)))
1115
file_ie = self._byid[file_id]
1116
old_parent = self._byid[file_ie.parent_id]
1118
# TODO: Don't leave things messed up if this fails
1120
del old_parent.children[file_ie.name]
1121
new_parent.children[new_name] = file_ie
1123
file_ie.name = new_name
1124
file_ie.parent_id = new_parent_id
1131
def is_valid_name(name):
1133
if _NAME_RE == None:
1134
_NAME_RE = re.compile(r'^[^/\\]+$')
1136
return bool(_NAME_RE.match(name))
added
removed
bzrlib/inventory.py
