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.
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}' %
353
"""Clone this inventory entry."""
354
raise NotImplementedError
356
def _describe_snapshot_change(self, previous_entries):
357
"""Describe how this entry will have changed in a new commit.
359
:param previous_entries: Dictionary from revision_id to inventory entry.
361
:returns: One-word description: "merged", "added", "renamed", "modified".
363
# XXX: This assumes that the file *has* changed -- it should probably
364
# be fused with whatever does that detection. Why not just a single
365
# thing to compare the entries?
367
# TODO: Return some kind of object describing all the possible
368
# dimensions that can change, not just a string. That can then give
369
# both old and new names for renames, etc.
371
if len(previous_entries) > 1:
373
elif len(previous_entries) == 0:
375
the_parent, = previous_entries.values()
376
if self.parent_id != the_parent.parent_id:
377
# actually, moved to another directory
379
elif self.name != the_parent.name:
384
return ("%s(%r, %r, parent_id=%r)"
385
% (self.__class__.__name__,
390
def snapshot(self, revision, path, previous_entries,
391
work_tree, weave_store, transaction):
392
"""Make a snapshot of this entry which may or may not have changed.
394
This means that all its fields are populated, that it has its
395
text stored in the text store or weave.
397
mutter('new parents of %s are %r', path, previous_entries)
398
self._read_tree_state(path, work_tree)
399
if len(previous_entries) == 1:
400
# cannot be unchanged unless there is only one parent file rev.
401
parent_ie = previous_entries.values()[0]
402
if self._unchanged(parent_ie):
403
mutter("found unchanged entry")
404
self.revision = parent_ie.revision
406
return self._snapshot_into_revision(revision, previous_entries,
407
work_tree, weave_store, transaction)
409
def _snapshot_into_revision(self, revision, previous_entries, work_tree,
410
weave_store, transaction):
411
"""Record this revision unconditionally into a store.
413
The entry's last-changed revision property (`revision`) is updated to
414
that of the new revision.
416
:param revision: id of the new revision that is being recorded.
418
:returns: String description of the commit (e.g. "merged", "modified"), etc.
420
mutter('new revision {%s} for {%s}', revision, self.file_id)
421
self.revision = revision
422
change = self._describe_snapshot_change(previous_entries)
423
self._snapshot_text(previous_entries, work_tree, weave_store,
427
def _snapshot_text(self, file_parents, work_tree, weave_store, transaction):
428
"""Record the 'text' of this entry, whatever form that takes.
430
This default implementation simply adds an empty text.
432
mutter('storing file {%s} in revision {%s}',
433
self.file_id, self.revision)
434
self._add_text_to_weave([], file_parents.keys(), weave_store, transaction)
436
def __eq__(self, other):
437
if not isinstance(other, InventoryEntry):
438
return NotImplemented
440
return ((self.file_id == other.file_id)
441
and (self.name == other.name)
442
and (other.symlink_target == self.symlink_target)
443
and (self.text_sha1 == other.text_sha1)
444
and (self.text_size == other.text_size)
445
and (self.text_id == other.text_id)
446
and (self.parent_id == other.parent_id)
447
and (self.kind == other.kind)
448
and (self.revision == other.revision)
449
and (self.executable == other.executable)
452
def __ne__(self, other):
453
return not (self == other)
456
raise ValueError('not hashable')
458
def _unchanged(self, previous_ie):
459
"""Has this entry changed relative to previous_ie.
461
This method should be overriden in child classes.
464
# different inv parent
465
if previous_ie.parent_id != self.parent_id:
468
elif previous_ie.name != self.name:
472
def _read_tree_state(self, path, work_tree):
473
"""Populate fields in the inventory entry from the given tree.
475
Note that this should be modified to be a noop on virtual trees
476
as all entries created there are prepopulated.
478
# TODO: Rather than running this manually, we should check the
479
# working sha1 and other expensive properties when they're
480
# first requested, or preload them if they're already known
481
pass # nothing to do by default
483
def _forget_tree_state(self):
487
class RootEntry(InventoryEntry):
489
def _check(self, checker, rev_id, tree):
490
"""See InventoryEntry._check"""
492
def __init__(self, file_id):
493
self.file_id = file_id
495
self.kind = 'root_directory'
496
self.parent_id = None
499
def __eq__(self, other):
500
if not isinstance(other, RootEntry):
501
return NotImplemented
503
return (self.file_id == other.file_id) \
504
and (self.children == other.children)
507
class InventoryDirectory(InventoryEntry):
508
"""A directory in an inventory."""
510
def _check(self, checker, rev_id, tree):
511
"""See InventoryEntry._check"""
512
if self.text_sha1 != None or self.text_size != None or self.text_id != None:
513
raise BzrCheckError('directory {%s} has text in revision {%s}'
514
% (self.file_id, rev_id))
517
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
518
other.revision = self.revision
519
# note that children are *not* copied; they're pulled across when
523
def __init__(self, file_id, name, parent_id):
524
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
526
self.kind = 'directory'
528
def kind_character(self):
529
"""See InventoryEntry.kind_character."""
532
def _put_in_tar(self, item, tree):
533
"""See InventoryEntry._put_in_tar."""
534
item.type = tarfile.DIRTYPE
541
def _put_on_disk(self, fullpath, tree):
542
"""See InventoryEntry._put_on_disk."""
546
class InventoryFile(InventoryEntry):
547
"""A file in an inventory."""
549
def _check(self, checker, tree_revision_id, tree):
550
"""See InventoryEntry._check"""
551
t = (self.file_id, self.revision)
552
if t in checker.checked_texts:
553
prev_sha = checker.checked_texts[t]
554
if prev_sha != self.text_sha1:
555
raise BzrCheckError('mismatched sha1 on {%s} in {%s}' %
556
(self.file_id, tree_revision_id))
558
checker.repeated_text_cnt += 1
561
if self.file_id not in checker.checked_weaves:
562
mutter('check weave {%s}', self.file_id)
563
w = tree.get_weave(self.file_id)
564
# Not passing a progress bar, because it creates a new
565
# progress, which overwrites the current progress,
566
# and doesn't look nice
568
checker.checked_weaves[self.file_id] = True
570
w = tree.get_weave(self.file_id)
572
mutter('check version {%s} of {%s}', tree_revision_id, self.file_id)
573
checker.checked_text_cnt += 1
574
# We can't check the length, because Weave doesn't store that
575
# information, and the whole point of looking at the weave's
576
# sha1sum is that we don't have to extract the text.
577
if self.text_sha1 != w.get_sha1(self.revision):
578
raise BzrCheckError('text {%s} version {%s} wrong sha1'
579
% (self.file_id, self.revision))
580
checker.checked_texts[t] = self.text_sha1
583
other = InventoryFile(self.file_id, self.name, self.parent_id)
584
other.executable = self.executable
585
other.text_id = self.text_id
586
other.text_sha1 = self.text_sha1
587
other.text_size = self.text_size
588
other.revision = self.revision
591
def detect_changes(self, old_entry):
592
"""See InventoryEntry.detect_changes."""
593
assert self.text_sha1 != None
594
assert old_entry.text_sha1 != None
595
text_modified = (self.text_sha1 != old_entry.text_sha1)
596
meta_modified = (self.executable != old_entry.executable)
597
return text_modified, meta_modified
599
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
600
output_to, reverse=False):
601
"""See InventoryEntry._diff."""
603
from_text = tree.get_file(self.file_id).readlines()
605
to_text = to_tree.get_file(to_entry.file_id).readlines()
609
text_diff(from_label, from_text,
610
to_label, to_text, output_to)
612
text_diff(to_label, to_text,
613
from_label, from_text, output_to)
616
label_pair = (to_label, from_label)
618
label_pair = (from_label, to_label)
619
print >> output_to, "Binary files %s and %s differ" % label_pair
622
"""See InventoryEntry.has_text."""
625
def __init__(self, file_id, name, parent_id):
626
super(InventoryFile, self).__init__(file_id, name, parent_id)
629
def kind_character(self):
630
"""See InventoryEntry.kind_character."""
633
def _put_in_tar(self, item, tree):
634
"""See InventoryEntry._put_in_tar."""
635
item.type = tarfile.REGTYPE
636
fileobj = tree.get_file(self.file_id)
637
item.size = self.text_size
638
if tree.is_executable(self.file_id):
644
def _put_on_disk(self, fullpath, tree):
645
"""See InventoryEntry._put_on_disk."""
646
pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
647
if tree.is_executable(self.file_id):
648
os.chmod(fullpath, 0755)
650
def _read_tree_state(self, path, work_tree):
651
"""See InventoryEntry._read_tree_state."""
652
self.text_sha1 = work_tree.get_file_sha1(self.file_id)
653
self.executable = work_tree.is_executable(self.file_id)
655
def _forget_tree_state(self):
656
self.text_sha1 = None
657
self.executable = None
659
def _snapshot_text(self, file_parents, work_tree, weave_store, transaction):
660
"""See InventoryEntry._snapshot_text."""
661
mutter('storing file {%s} in revision {%s}',
662
self.file_id, self.revision)
663
# special case to avoid diffing on renames or
665
if (len(file_parents) == 1
666
and self.text_sha1 == file_parents.values()[0].text_sha1
667
and self.text_size == file_parents.values()[0].text_size):
668
previous_ie = file_parents.values()[0]
669
versionedfile = weave_store.get_weave(self.file_id, transaction)
670
versionedfile.clone_text(self.revision, previous_ie.revision, file_parents.keys())
672
new_lines = work_tree.get_file(self.file_id).readlines()
673
self._add_text_to_weave(new_lines, file_parents.keys(), weave_store,
675
self.text_sha1 = sha_strings(new_lines)
676
self.text_size = sum(map(len, new_lines))
679
def _unchanged(self, previous_ie):
680
"""See InventoryEntry._unchanged."""
681
compatible = super(InventoryFile, self)._unchanged(previous_ie)
682
if self.text_sha1 != previous_ie.text_sha1:
685
# FIXME: 20050930 probe for the text size when getting sha1
686
# in _read_tree_state
687
self.text_size = previous_ie.text_size
688
if self.executable != previous_ie.executable:
693
class InventoryLink(InventoryEntry):
694
"""A file in an inventory."""
696
__slots__ = ['symlink_target']
698
def _check(self, checker, rev_id, tree):
699
"""See InventoryEntry._check"""
700
if self.text_sha1 != None or self.text_size != None or self.text_id != None:
701
raise BzrCheckError('symlink {%s} has text in revision {%s}'
702
% (self.file_id, rev_id))
703
if self.symlink_target == None:
704
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
705
% (self.file_id, rev_id))
708
other = InventoryLink(self.file_id, self.name, self.parent_id)
709
other.symlink_target = self.symlink_target
710
other.revision = self.revision
713
def detect_changes(self, old_entry):
714
"""See InventoryEntry.detect_changes."""
715
# FIXME: which _modified field should we use ? RBC 20051003
716
text_modified = (self.symlink_target != old_entry.symlink_target)
718
mutter(" symlink target changed")
719
meta_modified = False
720
return text_modified, meta_modified
722
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
723
output_to, reverse=False):
724
"""See InventoryEntry._diff."""
725
from_text = self.symlink_target
726
if to_entry is not None:
727
to_text = to_entry.symlink_target
732
print >>output_to, '=== target changed %r => %r' % (from_text, to_text)
735
print >>output_to, '=== target was %r' % self.symlink_target
737
print >>output_to, '=== target is %r' % self.symlink_target
739
def __init__(self, file_id, name, parent_id):
740
super(InventoryLink, self).__init__(file_id, name, parent_id)
741
self.kind = 'symlink'
743
def kind_character(self):
744
"""See InventoryEntry.kind_character."""
747
def _put_in_tar(self, item, tree):
748
"""See InventoryEntry._put_in_tar."""
749
item.type = tarfile.SYMTYPE
753
item.linkname = self.symlink_target
756
def _put_on_disk(self, fullpath, tree):
757
"""See InventoryEntry._put_on_disk."""
759
os.symlink(self.symlink_target, fullpath)
761
raise BzrError("Failed to create symlink %r -> %r, error: %s" % (fullpath, self.symlink_target, e))
763
def _read_tree_state(self, path, work_tree):
764
"""See InventoryEntry._read_tree_state."""
765
self.symlink_target = work_tree.get_symlink_target(self.file_id)
767
def _forget_tree_state(self):
768
self.symlink_target = None
770
def _unchanged(self, previous_ie):
771
"""See InventoryEntry._unchanged."""
772
compatible = super(InventoryLink, self)._unchanged(previous_ie)
773
if self.symlink_target != previous_ie.symlink_target:
778
class Inventory(object):
779
"""Inventory of versioned files in a tree.
781
This describes which file_id is present at each point in the tree,
782
and possibly the SHA-1 or other information about the file.
783
Entries can be looked up either by path or by file_id.
785
The inventory represents a typical unix file tree, with
786
directories containing files and subdirectories. We never store
787
the full path to a file, because renaming a directory implicitly
788
moves all of its contents. This class internally maintains a
789
lookup tree that allows the children under a directory to be
792
InventoryEntry objects must not be modified after they are
793
inserted, other than through the Inventory API.
795
>>> inv = Inventory()
796
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
797
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT')
798
>>> inv['123-123'].name
801
May be treated as an iterator or set to look up file ids:
803
>>> bool(inv.path2id('hello.c'))
808
May also look up by name:
810
>>> [x[0] for x in inv.iter_entries()]
812
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
813
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
814
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT-12345678-12345678')
816
def __init__(self, root_id=ROOT_ID, revision_id=None):
817
"""Create or read an inventory.
819
If a working directory is specified, the inventory is read
820
from there. If the file is specified, read from that. If not,
821
the inventory is created empty.
823
The inventory is created with a default root directory, with
826
# We are letting Branch.create() create a unique inventory
827
# root id. Rather than generating a random one here.
829
# root_id = bzrlib.branch.gen_file_id('TREE_ROOT')
830
self.root = RootEntry(root_id)
831
self.revision_id = revision_id
832
self._byid = {self.root.file_id: self.root}
836
# TODO: jam 20051218 Should copy also copy the revision_id?
837
other = Inventory(self.root.file_id)
838
# copy recursively so we know directories will be added before
839
# their children. There are more efficient ways than this...
840
for path, entry in self.iter_entries():
841
if entry == self.root:
843
other.add(entry.copy())
848
return iter(self._byid)
852
"""Returns number of entries."""
853
return len(self._byid)
856
def iter_entries(self, from_dir=None):
857
"""Return (path, entry) pairs, in order by name."""
861
elif isinstance(from_dir, basestring):
862
from_dir = self._byid[from_dir]
864
kids = from_dir.children.items()
866
for name, ie in kids:
868
if ie.kind == 'directory':
869
for cn, cie in self.iter_entries(from_dir=ie.file_id):
870
yield pathjoin(name, cn), cie
874
"""Return list of (path, ie) for all entries except the root.
876
This may be faster than iter_entries.
879
def descend(dir_ie, dir_path):
880
kids = dir_ie.children.items()
882
for name, ie in kids:
883
child_path = pathjoin(dir_path, name)
884
accum.append((child_path, ie))
885
if ie.kind == 'directory':
886
descend(ie, child_path)
888
descend(self.root, u'')
892
def directories(self):
893
"""Return (path, entry) pairs for all directories, including the root.
896
def descend(parent_ie, parent_path):
897
accum.append((parent_path, parent_ie))
899
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
902
for name, child_ie in kids:
903
child_path = pathjoin(parent_path, name)
904
descend(child_ie, child_path)
905
descend(self.root, u'')
910
def __contains__(self, file_id):
911
"""True if this entry contains a file with given id.
913
>>> inv = Inventory()
914
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
915
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT')
921
return file_id in self._byid
924
def __getitem__(self, file_id):
925
"""Return the entry for given file_id.
927
>>> inv = Inventory()
928
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
929
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT')
930
>>> inv['123123'].name
934
return self._byid[file_id]
937
raise BzrError("can't look up file_id None")
939
raise BzrError("file_id {%s} not in inventory" % file_id)
942
def get_file_kind(self, file_id):
943
return self._byid[file_id].kind
945
def get_child(self, parent_id, filename):
946
return self[parent_id].children.get(filename)
949
def add(self, entry):
950
"""Add entry to inventory.
952
To add a file to a branch ready to be committed, use Branch.add,
955
Returns the new entry object.
957
if entry.file_id in self._byid:
958
raise BzrError("inventory already contains entry with id {%s}" % entry.file_id)
960
if entry.parent_id == ROOT_ID or entry.parent_id is None:
961
entry.parent_id = self.root.file_id
964
parent = self._byid[entry.parent_id]
966
raise BzrError("parent_id {%s} not in inventory" % entry.parent_id)
968
if parent.children.has_key(entry.name):
969
raise BzrError("%s is already versioned" %
970
pathjoin(self.id2path(parent.file_id), entry.name))
972
self._byid[entry.file_id] = entry
973
parent.children[entry.name] = entry
977
def add_path(self, relpath, kind, file_id=None):
978
"""Add entry from a path.
980
The immediate parent must already be versioned.
982
Returns the new entry object."""
983
from bzrlib.workingtree import gen_file_id
985
parts = bzrlib.osutils.splitpath(relpath)
988
file_id = gen_file_id(relpath)
991
self.root = RootEntry(file_id)
992
self._byid = {self.root.file_id: self.root}
995
parent_path = parts[:-1]
996
parent_id = self.path2id(parent_path)
997
if parent_id == None:
998
raise NotVersionedError(path=parent_path)
999
if kind == 'directory':
1000
ie = InventoryDirectory(file_id, parts[-1], parent_id)
1001
elif kind == 'file':
1002
ie = InventoryFile(file_id, parts[-1], parent_id)
1003
elif kind == 'symlink':
1004
ie = InventoryLink(file_id, parts[-1], parent_id)
1006
raise BzrError("unknown kind %r" % kind)
1010
def __delitem__(self, file_id):
1011
"""Remove entry by id.
1013
>>> inv = Inventory()
1014
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1015
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT')
1024
assert ie.parent_id is None or \
1025
self[ie.parent_id].children[ie.name] == ie
1027
del self._byid[file_id]
1028
if ie.parent_id is not None:
1029
del self[ie.parent_id].children[ie.name]
1032
def __eq__(self, other):
1033
"""Compare two sets by comparing their contents.
1035
>>> i1 = Inventory()
1036
>>> i2 = Inventory()
1039
>>> i1.add(InventoryFile('123', 'foo', ROOT_ID))
1040
InventoryFile('123', 'foo', parent_id='TREE_ROOT')
1043
>>> i2.add(InventoryFile('123', 'foo', ROOT_ID))
1044
InventoryFile('123', 'foo', parent_id='TREE_ROOT')
1048
if not isinstance(other, Inventory):
1049
return NotImplemented
1051
if len(self._byid) != len(other._byid):
1052
# shortcut: obviously not the same
1055
return self._byid == other._byid
1058
def __ne__(self, other):
1059
return not self.__eq__(other)
1063
raise ValueError('not hashable')
1065
def _iter_file_id_parents(self, file_id):
1066
"""Yield the parents of file_id up to the root."""
1067
while file_id != None:
1069
ie = self._byid[file_id]
1071
raise BzrError("file_id {%s} not found in inventory" % file_id)
1073
file_id = ie.parent_id
1075
def get_idpath(self, file_id):
1076
"""Return a list of file_ids for the path to an entry.
1078
The list contains one element for each directory followed by
1079
the id of the file itself. So the length of the returned list
1080
is equal to the depth of the file in the tree, counting the
1081
root directory as depth 1.
1084
for parent in self._iter_file_id_parents(file_id):
1085
p.insert(0, parent.file_id)
1088
def id2path(self, file_id):
1089
"""Return as a string the path to file_id.
1092
>>> e = i.add(InventoryDirectory('src-id', 'src', ROOT_ID))
1093
>>> e = i.add(InventoryFile('foo-id', 'foo.c', parent_id='src-id'))
1094
>>> print i.id2path('foo-id')
1097
# get all names, skipping root
1098
return '/'.join(reversed(
1099
[parent.name for parent in
1100
self._iter_file_id_parents(file_id)][:-1]))
1102
def path2id(self, name):
1103
"""Walk down through directories to return entry of last component.
1105
names may be either a list of path components, or a single
1106
string, in which case it is automatically split.
1108
This returns the entry of the last component in the path,
1109
which may be either a file or a directory.
1111
Returns None iff the path is not found.
1113
if isinstance(name, types.StringTypes):
1114
name = splitpath(name)
1116
mutter("lookup path %r" % name)
1121
cie = parent.children[f]
1122
assert cie.name == f
1123
assert cie.parent_id == parent.file_id
1126
# or raise an error?
1129
return parent.file_id
1132
def has_filename(self, names):
1133
return bool(self.path2id(names))
1136
def has_id(self, file_id):
1137
return self._byid.has_key(file_id)
1140
def rename(self, file_id, new_parent_id, new_name):
1141
"""Move a file within the inventory.
1143
This can change either the name, or the parent, or both.
1145
This does not move the working file."""
1146
if not is_valid_name(new_name):
1147
raise BzrError("not an acceptable filename: %r" % new_name)
1149
new_parent = self._byid[new_parent_id]
1150
if new_name in new_parent.children:
1151
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
1153
new_parent_idpath = self.get_idpath(new_parent_id)
1154
if file_id in new_parent_idpath:
1155
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
1156
% (self.id2path(file_id), self.id2path(new_parent_id)))
1158
file_ie = self._byid[file_id]
1159
old_parent = self._byid[file_ie.parent_id]
1161
# TODO: Don't leave things messed up if this fails
1163
del old_parent.children[file_ie.name]
1164
new_parent.children[new_name] = file_ie
1166
file_ie.name = new_name
1167
file_ie.parent_id = new_parent_id
1174
def is_valid_name(name):
1176
if _NAME_RE == None:
1177
_NAME_RE = re.compile(r'^[^/\\]+$')
1179
return bool(_NAME_RE.match(name))
added
removed
bzrlib/inventory.py
