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.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
:param checker: Check object providing context for the checks;
333
can be used to find out what parts of the repository have already
335
:param rev_id: Revision id from which this InventoryEntry was loaded.
336
Not necessarily the last-changed revision for this file.
337
:param inv: Inventory from which the entry was loaded.
338
:param tree: RevisionTree for this entry.
340
if self.parent_id != None:
341
if not inv.has_id(self.parent_id):
342
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
343
% (self.parent_id, rev_id))
344
self._check(checker, rev_id, tree)
346
def _check(self, checker, rev_id, tree):
347
"""Check this inventory entry for kind specific errors."""
348
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
352
"""Clone this inventory entry."""
353
raise NotImplementedError
355
def _describe_snapshot_change(self, previous_entries):
356
"""Describe how this entry will have changed in a new commit.
358
:param previous_entries: Dictionary from revision_id to inventory entry.
360
:returns: One-word description: "merged", "added", "renamed", "modified".
362
# XXX: This assumes that the file *has* changed -- it should probably
363
# be fused with whatever does that detection. Why not just a single
364
# thing to compare the entries?
366
# TODO: Return some kind of object describing all the possible
367
# dimensions that can change, not just a string. That can then give
368
# both old and new names for renames, etc.
370
if len(previous_entries) > 1:
372
elif len(previous_entries) == 0:
374
the_parent, = previous_entries.values()
375
if self.parent_id != the_parent.parent_id:
376
# actually, moved to another directory
378
elif self.name != the_parent.name:
383
return ("%s(%r, %r, parent_id=%r)"
384
% (self.__class__.__name__,
389
def snapshot(self, revision, path, previous_entries,
390
work_tree, weave_store, transaction):
391
"""Make a snapshot of this entry which may or may not have changed.
393
This means that all its fields are populated, that it has its
394
text stored in the text store or weave.
396
mutter('new parents of %s are %r', path, previous_entries)
397
self._read_tree_state(path, work_tree)
398
if len(previous_entries) == 1:
399
# cannot be unchanged unless there is only one parent file rev.
400
parent_ie = previous_entries.values()[0]
401
if self._unchanged(parent_ie):
402
mutter("found unchanged entry")
403
self.revision = parent_ie.revision
405
return self._snapshot_into_revision(revision, previous_entries,
406
work_tree, weave_store, transaction)
408
def _snapshot_into_revision(self, revision, previous_entries, work_tree,
409
weave_store, transaction):
410
"""Record this revision unconditionally into a store.
412
The entry's last-changed revision property (`revision`) is updated to
413
that of the new revision.
415
:param revision: id of the new revision that is being recorded.
417
:returns: String description of the commit (e.g. "merged", "modified"), etc.
419
mutter('new revision {%s} for {%s}', revision, self.file_id)
420
self.revision = revision
421
change = self._describe_snapshot_change(previous_entries)
422
self._snapshot_text(previous_entries, work_tree, weave_store,
426
def _snapshot_text(self, file_parents, work_tree, weave_store, transaction):
427
"""Record the 'text' of this entry, whatever form that takes.
429
This default implementation simply adds an empty text.
431
mutter('storing file {%s} in revision {%s}',
432
self.file_id, self.revision)
433
self._add_text_to_weave([], file_parents.keys(), weave_store, transaction)
435
def __eq__(self, other):
436
if not isinstance(other, InventoryEntry):
437
return NotImplemented
439
return ((self.file_id == other.file_id)
440
and (self.name == other.name)
441
and (other.symlink_target == self.symlink_target)
442
and (self.text_sha1 == other.text_sha1)
443
and (self.text_size == other.text_size)
444
and (self.text_id == other.text_id)
445
and (self.parent_id == other.parent_id)
446
and (self.kind == other.kind)
447
and (self.revision == other.revision)
448
and (self.executable == other.executable)
451
def __ne__(self, other):
452
return not (self == other)
455
raise ValueError('not hashable')
457
def _unchanged(self, previous_ie):
458
"""Has this entry changed relative to previous_ie.
460
This method should be overriden in child classes.
463
# different inv parent
464
if previous_ie.parent_id != self.parent_id:
467
elif previous_ie.name != self.name:
471
def _read_tree_state(self, path, work_tree):
472
"""Populate fields in the inventory entry from the given tree.
474
Note that this should be modified to be a noop on virtual trees
475
as all entries created there are prepopulated.
477
# TODO: Rather than running this manually, we should check the
478
# working sha1 and other expensive properties when they're
479
# first requested, or preload them if they're already known
480
pass # nothing to do by default
482
def _forget_tree_state(self):
486
class RootEntry(InventoryEntry):
488
def _check(self, checker, rev_id, tree):
489
"""See InventoryEntry._check"""
491
def __init__(self, file_id):
492
self.file_id = file_id
494
self.kind = 'root_directory'
495
self.parent_id = None
498
def __eq__(self, other):
499
if not isinstance(other, RootEntry):
500
return NotImplemented
502
return (self.file_id == other.file_id) \
503
and (self.children == other.children)
506
class InventoryDirectory(InventoryEntry):
507
"""A directory in an inventory."""
509
def _check(self, checker, rev_id, tree):
510
"""See InventoryEntry._check"""
511
if self.text_sha1 != None or self.text_size != None or self.text_id != None:
512
raise BzrCheckError('directory {%s} has text in revision {%s}'
513
% (self.file_id, rev_id))
516
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
517
other.revision = self.revision
518
# note that children are *not* copied; they're pulled across when
522
def __init__(self, file_id, name, parent_id):
523
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
525
self.kind = 'directory'
527
def kind_character(self):
528
"""See InventoryEntry.kind_character."""
531
def _put_in_tar(self, item, tree):
532
"""See InventoryEntry._put_in_tar."""
533
item.type = tarfile.DIRTYPE
540
def _put_on_disk(self, fullpath, tree):
541
"""See InventoryEntry._put_on_disk."""
545
class InventoryFile(InventoryEntry):
546
"""A file in an inventory."""
548
def _check(self, checker, tree_revision_id, tree):
549
"""See InventoryEntry._check"""
550
t = (self.file_id, self.revision)
551
if t in checker.checked_texts:
552
prev_sha = checker.checked_texts[t]
553
if prev_sha != self.text_sha1:
554
raise BzrCheckError('mismatched sha1 on {%s} in {%s}' %
555
(self.file_id, tree_revision_id))
557
checker.repeated_text_cnt += 1
560
if self.file_id not in checker.checked_weaves:
561
mutter('check weave {%s}', self.file_id)
562
w = tree.get_weave(self.file_id)
563
# Not passing a progress bar, because it creates a new
564
# progress, which overwrites the current progress,
565
# and doesn't look nice
567
checker.checked_weaves[self.file_id] = True
569
w = tree.get_weave(self.file_id)
571
mutter('check version {%s} of {%s}', tree_revision_id, self.file_id)
572
checker.checked_text_cnt += 1
573
# We can't check the length, because Weave doesn't store that
574
# information, and the whole point of looking at the weave's
575
# sha1sum is that we don't have to extract the text.
576
if self.text_sha1 != w.get_sha1(self.revision):
577
raise BzrCheckError('text {%s} version {%s} wrong sha1'
578
% (self.file_id, self.revision))
579
checker.checked_texts[t] = self.text_sha1
582
other = InventoryFile(self.file_id, self.name, self.parent_id)
583
other.executable = self.executable
584
other.text_id = self.text_id
585
other.text_sha1 = self.text_sha1
586
other.text_size = self.text_size
587
other.revision = self.revision
590
def detect_changes(self, old_entry):
591
"""See InventoryEntry.detect_changes."""
592
assert self.text_sha1 != None
593
assert old_entry.text_sha1 != None
594
text_modified = (self.text_sha1 != old_entry.text_sha1)
595
meta_modified = (self.executable != old_entry.executable)
596
return text_modified, meta_modified
598
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
599
output_to, reverse=False):
600
"""See InventoryEntry._diff."""
601
from_text = tree.get_file(self.file_id).readlines()
603
to_text = to_tree.get_file(to_entry.file_id).readlines()
607
text_diff(from_label, from_text,
608
to_label, to_text, output_to)
610
text_diff(to_label, to_text,
611
from_label, from_text, output_to)
614
"""See InventoryEntry.has_text."""
617
def __init__(self, file_id, name, parent_id):
618
super(InventoryFile, self).__init__(file_id, name, parent_id)
621
def kind_character(self):
622
"""See InventoryEntry.kind_character."""
625
def _put_in_tar(self, item, tree):
626
"""See InventoryEntry._put_in_tar."""
627
item.type = tarfile.REGTYPE
628
fileobj = tree.get_file(self.file_id)
629
item.size = self.text_size
630
if tree.is_executable(self.file_id):
636
def _put_on_disk(self, fullpath, tree):
637
"""See InventoryEntry._put_on_disk."""
638
pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
639
if tree.is_executable(self.file_id):
640
os.chmod(fullpath, 0755)
642
def _read_tree_state(self, path, work_tree):
643
"""See InventoryEntry._read_tree_state."""
644
self.text_sha1 = work_tree.get_file_sha1(self.file_id)
645
self.executable = work_tree.is_executable(self.file_id)
647
def _forget_tree_state(self):
648
self.text_sha1 = None
649
self.executable = None
651
def _snapshot_text(self, file_parents, work_tree, weave_store, transaction):
652
"""See InventoryEntry._snapshot_text."""
653
mutter('storing file {%s} in revision {%s}',
654
self.file_id, self.revision)
655
# special case to avoid diffing on renames or
657
if (len(file_parents) == 1
658
and self.text_sha1 == file_parents.values()[0].text_sha1
659
and self.text_size == file_parents.values()[0].text_size):
660
previous_ie = file_parents.values()[0]
661
versionedfile = weave_store.get_weave(self.file_id, transaction)
662
versionedfile.clone_text(self.revision, previous_ie.revision, file_parents.keys())
664
new_lines = work_tree.get_file(self.file_id).readlines()
665
self._add_text_to_weave(new_lines, file_parents.keys(), weave_store,
667
self.text_sha1 = sha_strings(new_lines)
668
self.text_size = sum(map(len, new_lines))
671
def _unchanged(self, previous_ie):
672
"""See InventoryEntry._unchanged."""
673
compatible = super(InventoryFile, self)._unchanged(previous_ie)
674
if self.text_sha1 != previous_ie.text_sha1:
677
# FIXME: 20050930 probe for the text size when getting sha1
678
# in _read_tree_state
679
self.text_size = previous_ie.text_size
680
if self.executable != previous_ie.executable:
685
class InventoryLink(InventoryEntry):
686
"""A file in an inventory."""
688
__slots__ = ['symlink_target']
690
def _check(self, checker, rev_id, tree):
691
"""See InventoryEntry._check"""
692
if self.text_sha1 != None or self.text_size != None or self.text_id != None:
693
raise BzrCheckError('symlink {%s} has text in revision {%s}'
694
% (self.file_id, rev_id))
695
if self.symlink_target == None:
696
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
697
% (self.file_id, rev_id))
700
other = InventoryLink(self.file_id, self.name, self.parent_id)
701
other.symlink_target = self.symlink_target
702
other.revision = self.revision
705
def detect_changes(self, old_entry):
706
"""See InventoryEntry.detect_changes."""
707
# FIXME: which _modified field should we use ? RBC 20051003
708
text_modified = (self.symlink_target != old_entry.symlink_target)
710
mutter(" symlink target changed")
711
meta_modified = False
712
return text_modified, meta_modified
714
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
715
output_to, reverse=False):
716
"""See InventoryEntry._diff."""
717
from_text = self.symlink_target
718
if to_entry is not None:
719
to_text = to_entry.symlink_target
724
print >>output_to, '=== target changed %r => %r' % (from_text, to_text)
727
print >>output_to, '=== target was %r' % self.symlink_target
729
print >>output_to, '=== target is %r' % self.symlink_target
731
def __init__(self, file_id, name, parent_id):
732
super(InventoryLink, self).__init__(file_id, name, parent_id)
733
self.kind = 'symlink'
735
def kind_character(self):
736
"""See InventoryEntry.kind_character."""
739
def _put_in_tar(self, item, tree):
740
"""See InventoryEntry._put_in_tar."""
741
item.type = tarfile.SYMTYPE
745
item.linkname = self.symlink_target
748
def _put_on_disk(self, fullpath, tree):
749
"""See InventoryEntry._put_on_disk."""
751
os.symlink(self.symlink_target, fullpath)
753
raise BzrError("Failed to create symlink %r -> %r, error: %s" % (fullpath, self.symlink_target, e))
755
def _read_tree_state(self, path, work_tree):
756
"""See InventoryEntry._read_tree_state."""
757
self.symlink_target = work_tree.get_symlink_target(self.file_id)
759
def _forget_tree_state(self):
760
self.symlink_target = None
762
def _unchanged(self, previous_ie):
763
"""See InventoryEntry._unchanged."""
764
compatible = super(InventoryLink, self)._unchanged(previous_ie)
765
if self.symlink_target != previous_ie.symlink_target:
770
class Inventory(object):
771
"""Inventory of versioned files in a tree.
773
This describes which file_id is present at each point in the tree,
774
and possibly the SHA-1 or other information about the file.
775
Entries can be looked up either by path or by file_id.
777
The inventory represents a typical unix file tree, with
778
directories containing files and subdirectories. We never store
779
the full path to a file, because renaming a directory implicitly
780
moves all of its contents. This class internally maintains a
781
lookup tree that allows the children under a directory to be
784
InventoryEntry objects must not be modified after they are
785
inserted, other than through the Inventory API.
787
>>> inv = Inventory()
788
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
789
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT')
790
>>> inv['123-123'].name
793
May be treated as an iterator or set to look up file ids:
795
>>> bool(inv.path2id('hello.c'))
800
May also look up by name:
802
>>> [x[0] for x in inv.iter_entries()]
804
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
805
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
806
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT-12345678-12345678')
808
def __init__(self, root_id=ROOT_ID, revision_id=None):
809
"""Create or read an inventory.
811
If a working directory is specified, the inventory is read
812
from there. If the file is specified, read from that. If not,
813
the inventory is created empty.
815
The inventory is created with a default root directory, with
818
# We are letting Branch.create() create a unique inventory
819
# root id. Rather than generating a random one here.
821
# root_id = bzrlib.branch.gen_file_id('TREE_ROOT')
822
self.root = RootEntry(root_id)
823
self.revision_id = revision_id
824
self._byid = {self.root.file_id: self.root}
828
# TODO: jam 20051218 Should copy also copy the revision_id?
829
other = Inventory(self.root.file_id)
830
# copy recursively so we know directories will be added before
831
# their children. There are more efficient ways than this...
832
for path, entry in self.iter_entries():
833
if entry == self.root:
835
other.add(entry.copy())
840
return iter(self._byid)
844
"""Returns number of entries."""
845
return len(self._byid)
848
def iter_entries(self, from_dir=None):
849
"""Return (path, entry) pairs, in order by name."""
853
elif isinstance(from_dir, basestring):
854
from_dir = self._byid[from_dir]
856
kids = from_dir.children.items()
858
for name, ie in kids:
860
if ie.kind == 'directory':
861
for cn, cie in self.iter_entries(from_dir=ie.file_id):
862
yield pathjoin(name, cn), cie
866
"""Return list of (path, ie) for all entries except the root.
868
This may be faster than iter_entries.
871
def descend(dir_ie, dir_path):
872
kids = dir_ie.children.items()
874
for name, ie in kids:
875
child_path = pathjoin(dir_path, name)
876
accum.append((child_path, ie))
877
if ie.kind == 'directory':
878
descend(ie, child_path)
880
descend(self.root, u'')
884
def directories(self):
885
"""Return (path, entry) pairs for all directories, including the root.
888
def descend(parent_ie, parent_path):
889
accum.append((parent_path, parent_ie))
891
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
894
for name, child_ie in kids:
895
child_path = pathjoin(parent_path, name)
896
descend(child_ie, child_path)
897
descend(self.root, u'')
902
def __contains__(self, file_id):
903
"""True if this entry contains a file with given id.
905
>>> inv = Inventory()
906
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
907
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT')
913
return file_id in self._byid
916
def __getitem__(self, file_id):
917
"""Return the entry for given file_id.
919
>>> inv = Inventory()
920
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
921
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT')
922
>>> inv['123123'].name
926
return self._byid[file_id]
929
raise BzrError("can't look up file_id None")
931
raise BzrError("file_id {%s} not in inventory" % file_id)
934
def get_file_kind(self, file_id):
935
return self._byid[file_id].kind
937
def get_child(self, parent_id, filename):
938
return self[parent_id].children.get(filename)
941
def add(self, entry):
942
"""Add entry to inventory.
944
To add a file to a branch ready to be committed, use Branch.add,
947
Returns the new entry object.
949
if entry.file_id in self._byid:
950
raise BzrError("inventory already contains entry with id {%s}" % entry.file_id)
952
if entry.parent_id == ROOT_ID or entry.parent_id is None:
953
entry.parent_id = self.root.file_id
956
parent = self._byid[entry.parent_id]
958
raise BzrError("parent_id {%s} not in inventory" % entry.parent_id)
960
if parent.children.has_key(entry.name):
961
raise BzrError("%s is already versioned" %
962
pathjoin(self.id2path(parent.file_id), entry.name))
964
self._byid[entry.file_id] = entry
965
parent.children[entry.name] = entry
969
def add_path(self, relpath, kind, file_id=None):
970
"""Add entry from a path.
972
The immediate parent must already be versioned.
974
Returns the new entry object."""
975
from bzrlib.workingtree import gen_file_id
977
parts = bzrlib.osutils.splitpath(relpath)
980
file_id = gen_file_id(relpath)
983
self.root = RootEntry(file_id)
984
self._byid = {self.root.file_id: self.root}
987
parent_path = parts[:-1]
988
parent_id = self.path2id(parent_path)
989
if parent_id == None:
990
raise NotVersionedError(path=parent_path)
991
if kind == 'directory':
992
ie = InventoryDirectory(file_id, parts[-1], parent_id)
994
ie = InventoryFile(file_id, parts[-1], parent_id)
995
elif kind == 'symlink':
996
ie = InventoryLink(file_id, parts[-1], parent_id)
998
raise BzrError("unknown kind %r" % kind)
1002
def __delitem__(self, file_id):
1003
"""Remove entry by id.
1005
>>> inv = Inventory()
1006
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1007
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT')
1016
assert ie.parent_id is None or \
1017
self[ie.parent_id].children[ie.name] == ie
1019
del self._byid[file_id]
1020
if ie.parent_id is not None:
1021
del self[ie.parent_id].children[ie.name]
1024
def __eq__(self, other):
1025
"""Compare two sets by comparing their contents.
1027
>>> i1 = Inventory()
1028
>>> i2 = Inventory()
1031
>>> i1.add(InventoryFile('123', 'foo', ROOT_ID))
1032
InventoryFile('123', 'foo', parent_id='TREE_ROOT')
1035
>>> i2.add(InventoryFile('123', 'foo', ROOT_ID))
1036
InventoryFile('123', 'foo', parent_id='TREE_ROOT')
1040
if not isinstance(other, Inventory):
1041
return NotImplemented
1043
if len(self._byid) != len(other._byid):
1044
# shortcut: obviously not the same
1047
return self._byid == other._byid
1050
def __ne__(self, other):
1051
return not self.__eq__(other)
1055
raise ValueError('not hashable')
1057
def _iter_file_id_parents(self, file_id):
1058
"""Yield the parents of file_id up to the root."""
1059
while file_id != None:
1061
ie = self._byid[file_id]
1063
raise BzrError("file_id {%s} not found in inventory" % file_id)
1065
file_id = ie.parent_id
1067
def get_idpath(self, file_id):
1068
"""Return a list of file_ids for the path to an entry.
1070
The list contains one element for each directory followed by
1071
the id of the file itself. So the length of the returned list
1072
is equal to the depth of the file in the tree, counting the
1073
root directory as depth 1.
1076
for parent in self._iter_file_id_parents(file_id):
1077
p.insert(0, parent.file_id)
1080
def id2path(self, file_id):
1081
"""Return as a string the path to file_id.
1084
>>> e = i.add(InventoryDirectory('src-id', 'src', ROOT_ID))
1085
>>> e = i.add(InventoryFile('foo-id', 'foo.c', parent_id='src-id'))
1086
>>> print i.id2path('foo-id')
1089
# get all names, skipping root
1090
return '/'.join(reversed(
1091
[parent.name for parent in
1092
self._iter_file_id_parents(file_id)][:-1]))
1094
def path2id(self, name):
1095
"""Walk down through directories to return entry of last component.
1097
names may be either a list of path components, or a single
1098
string, in which case it is automatically split.
1100
This returns the entry of the last component in the path,
1101
which may be either a file or a directory.
1103
Returns None iff the path is not found.
1105
if isinstance(name, types.StringTypes):
1106
name = splitpath(name)
1108
mutter("lookup path %r" % name)
1113
cie = parent.children[f]
1114
assert cie.name == f
1115
assert cie.parent_id == parent.file_id
1118
# or raise an error?
1121
return parent.file_id
1124
def has_filename(self, names):
1125
return bool(self.path2id(names))
1128
def has_id(self, file_id):
1129
return self._byid.has_key(file_id)
1132
def rename(self, file_id, new_parent_id, new_name):
1133
"""Move a file within the inventory.
1135
This can change either the name, or the parent, or both.
1137
This does not move the working file."""
1138
if not is_valid_name(new_name):
1139
raise BzrError("not an acceptable filename: %r" % new_name)
1141
new_parent = self._byid[new_parent_id]
1142
if new_name in new_parent.children:
1143
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
1145
new_parent_idpath = self.get_idpath(new_parent_id)
1146
if file_id in new_parent_idpath:
1147
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
1148
% (self.id2path(file_id), self.id2path(new_parent_id)))
1150
file_ie = self._byid[file_id]
1151
old_parent = self._byid[file_ie.parent_id]
1153
# TODO: Don't leave things messed up if this fails
1155
del old_parent.children[file_ie.name]
1156
new_parent.children[new_name] = file_ie
1158
file_ie.name = new_name
1159
file_ie.parent_id = new_parent_id
1166
def is_valid_name(name):
1168
if _NAME_RE == None:
1169
_NAME_RE = re.compile(r'^[^/\\]+$')
1171
return bool(_NAME_RE.match(name))
added
removed
bzrlib/inventory.py
