1
# Copyright (C) 2005, 2006 Canonical Ltd
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
# FIXME: This refactoring of the workingtree code doesn't seem to keep
18
# the WorkingTree's copy of the inventory in sync with the branch. The
19
# branch modifies its working inventory when it does a commit to make
20
# missing files permanently removed.
22
# TODO: Maybe also keep the full path of the entry, and the children?
23
# But those depend on its position within a particular inventory, and
24
# it would be nice not to need to hold the backpointer here.
26
# This should really be an id randomly assigned when the tree is
27
# created, but it's not for now.
39
from bzrlib.osutils import (pumpfile, quotefn, splitpath, joinpath,
40
pathjoin, sha_strings)
41
from bzrlib.errors import (NotVersionedError, InvalidEntryName,
42
BzrError, BzrCheckError, BinaryFile)
43
from bzrlib.trace import mutter
46
class InventoryEntry(object):
47
"""Description of a versioned file.
49
An InventoryEntry has the following fields, which are also
50
present in the XML inventory-entry element:
55
(within the parent directory)
58
file_id of the parent directory, or ROOT_ID
61
the revision_id in which this variation of this file was
65
Indicates that this file should be executable on systems
69
sha-1 of the text of the file
72
size in bytes of the text of the file
74
(reading a version 4 tree created a text_id field.)
79
>>> i.add(InventoryDirectory('123', 'src', ROOT_ID))
80
InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None)
81
>>> i.add(InventoryFile('2323', 'hello.c', parent_id='123'))
82
InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None)
83
>>> shouldbe = {0: 'src', 1: pathjoin('src','hello.c')}
84
>>> for ix, j in enumerate(i.iter_entries()):
85
... print (j[0] == shouldbe[ix], j[1])
87
(True, InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None))
88
(True, InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None))
89
>>> i.add(InventoryFile('2323', 'bye.c', '123'))
90
Traceback (most recent call last):
92
BzrError: inventory already contains entry with id {2323}
93
>>> i.add(InventoryFile('2324', 'bye.c', '123'))
94
InventoryFile('2324', 'bye.c', parent_id='123', sha1=None, len=None)
95
>>> i.add(InventoryDirectory('2325', 'wibble', '123'))
96
InventoryDirectory('2325', 'wibble', parent_id='123', revision=None)
97
>>> i.path2id('src/wibble')
101
>>> i.add(InventoryFile('2326', 'wibble.c', '2325'))
102
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
104
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
105
>>> for path, entry in i.iter_entries():
107
... assert i.path2id(path)
114
>>> i.id2path('2326')
115
'src/wibble/wibble.c'
118
# Constants returned by describe_change()
120
# TODO: These should probably move to some kind of FileChangeDescription
121
# class; that's like what's inside a TreeDelta but we want to be able to
122
# generate them just for one file at a time.
124
MODIFIED_AND_RENAMED = 'modified and renamed'
126
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
127
'text_id', 'parent_id', 'children', 'executable',
130
def detect_changes(self, old_entry):
131
"""Return a (text_modified, meta_modified) from this to old_entry.
133
_read_tree_state must have been called on self and old_entry prior to
134
calling detect_changes.
138
def diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
139
output_to, reverse=False):
140
"""Perform a diff from this to to_entry.
142
text_diff will be used for textual difference calculation.
143
This is a template method, override _diff in child classes.
145
self._read_tree_state(tree.id2path(self.file_id), tree)
147
# cannot diff from one kind to another - you must do a removal
148
# and an addif they do not match.
149
assert self.kind == to_entry.kind
150
to_entry._read_tree_state(to_tree.id2path(to_entry.file_id),
152
self._diff(text_diff, from_label, tree, to_label, to_entry, to_tree,
155
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
156
output_to, reverse=False):
157
"""Perform a diff between two entries of the same kind."""
159
def find_previous_heads(self, previous_inventories,
160
versioned_file_store,
163
"""Return the revisions and entries that directly precede this.
165
Returned as a map from revision to inventory entry.
167
This is a map containing the file revisions in all parents
168
for which the file exists, and its revision is not a parent of
169
any other. If the file is new, the set will be empty.
171
:param versioned_file_store: A store where ancestry data on this
172
file id can be queried.
173
:param transaction: The transaction that queries to the versioned
174
file store should be completed under.
175
:param entry_vf: The entry versioned file, if its already available.
177
def get_ancestors(weave, entry):
178
return set(weave.get_ancestry(entry.revision))
179
# revision:ie mapping for each ie found in previous_inventories.
181
# revision:ie mapping with one revision for each head.
183
# revision: ancestor list for each head
185
# identify candidate head revision ids.
186
for inv in previous_inventories:
187
if self.file_id in inv:
188
ie = inv[self.file_id]
189
assert ie.file_id == self.file_id
190
if ie.revision in candidates:
191
# same revision value in two different inventories:
192
# correct possible inconsistencies:
193
# * there was a bug in revision updates with 'x' bit
196
if candidates[ie.revision].executable != ie.executable:
197
candidates[ie.revision].executable = False
198
ie.executable = False
199
except AttributeError:
201
# must now be the same.
202
assert candidates[ie.revision] == ie
204
# add this revision as a candidate.
205
candidates[ie.revision] = ie
207
# common case optimisation
208
if len(candidates) == 1:
209
# if there is only one candidate revision found
210
# then we can opening the versioned file to access ancestry:
211
# there cannot be any ancestors to eliminate when there is
212
# only one revision available.
213
heads[ie.revision] = ie
216
# eliminate ancestors amongst the available candidates:
217
# heads are those that are not an ancestor of any other candidate
218
# - this provides convergence at a per-file level.
219
for ie in candidates.values():
220
# may be an ancestor of a known head:
221
already_present = 0 != len(
222
[head for head in heads
223
if ie.revision in head_ancestors[head]])
225
# an ancestor of an analyzed candidate.
227
# not an ancestor of a known head:
228
# load the versioned file for this file id if needed
230
entry_vf = versioned_file_store.get_weave_or_empty(
231
self.file_id, transaction)
232
ancestors = get_ancestors(entry_vf, ie)
233
# may knock something else out:
234
check_heads = list(heads.keys())
235
for head in check_heads:
236
if head in ancestors:
237
# this previously discovered 'head' is not
238
# really a head - its an ancestor of the newly
241
head_ancestors[ie.revision] = ancestors
242
heads[ie.revision] = ie
245
def get_tar_item(self, root, dp, now, tree):
246
"""Get a tarfile item and a file stream for its content."""
247
item = tarfile.TarInfo(pathjoin(root, dp))
248
# TODO: would be cool to actually set it to the timestamp of the
249
# revision it was last changed
251
fileobj = self._put_in_tar(item, tree)
255
"""Return true if the object this entry represents has textual data.
257
Note that textual data includes binary content.
259
Also note that all entries get weave files created for them.
260
This attribute is primarily used when upgrading from old trees that
261
did not have the weave index for all inventory entries.
265
def __init__(self, file_id, name, parent_id, text_id=None):
266
"""Create an InventoryEntry
268
The filename must be a single component, relative to the
269
parent directory; it cannot be a whole path or relative name.
271
>>> e = InventoryFile('123', 'hello.c', ROOT_ID)
276
>>> e = InventoryFile('123', 'src/hello.c', ROOT_ID)
277
Traceback (most recent call last):
278
InvalidEntryName: Invalid entry name: src/hello.c
280
assert isinstance(name, basestring), name
281
if '/' in name or '\\' in name:
282
raise InvalidEntryName(name=name)
283
self.executable = False
285
self.text_sha1 = None
286
self.text_size = None
287
self.file_id = file_id
289
self.text_id = text_id
290
self.parent_id = parent_id
291
self.symlink_target = None
293
def kind_character(self):
294
"""Return a short kind indicator useful for appending to names."""
295
raise BzrError('unknown kind %r' % self.kind)
297
known_kinds = ('file', 'directory', 'symlink', 'root_directory')
299
def _put_in_tar(self, item, tree):
300
"""populate item for stashing in a tar, and return the content stream.
302
If no content is available, return None.
304
raise BzrError("don't know how to export {%s} of kind %r" %
305
(self.file_id, self.kind))
307
def put_on_disk(self, dest, dp, tree):
308
"""Create a representation of self on disk in the prefix dest.
310
This is a template method - implement _put_on_disk in subclasses.
312
fullpath = pathjoin(dest, dp)
313
self._put_on_disk(fullpath, tree)
314
mutter(" export {%s} kind %s to %s", self.file_id,
317
def _put_on_disk(self, fullpath, tree):
318
"""Put this entry onto disk at fullpath, from tree tree."""
319
raise BzrError("don't know how to export {%s} of kind %r" % (self.file_id, self.kind))
321
def sorted_children(self):
322
l = self.children.items()
327
def versionable_kind(kind):
328
return kind in ('file', 'directory', 'symlink')
330
def check(self, checker, rev_id, inv, tree):
331
"""Check this inventory entry is intact.
333
This is a template method, override _check for kind specific
336
:param checker: Check object providing context for the checks;
337
can be used to find out what parts of the repository have already
339
:param rev_id: Revision id from which this InventoryEntry was loaded.
340
Not necessarily the last-changed revision for this file.
341
:param inv: Inventory from which the entry was loaded.
342
:param tree: RevisionTree for this entry.
344
if self.parent_id != None:
345
if not inv.has_id(self.parent_id):
346
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
347
% (self.parent_id, rev_id))
348
self._check(checker, rev_id, tree)
350
def _check(self, checker, rev_id, tree):
351
"""Check this inventory entry for kind specific errors."""
352
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
356
"""Clone this inventory entry."""
357
raise NotImplementedError
360
def describe_change(old_entry, new_entry):
361
"""Describe the change between old_entry and this.
363
This smells of being an InterInventoryEntry situation, but as its
364
the first one, we're making it a static method for now.
366
An entry with a different parent, or different name is considered
367
to be renamed. Reparenting is an internal detail.
368
Note that renaming the parent does not trigger a rename for the
371
# TODO: Perhaps return an object rather than just a string
372
if old_entry is new_entry:
373
# also the case of both being None
375
elif old_entry is None:
377
elif new_entry is None:
379
text_modified, meta_modified = new_entry.detect_changes(old_entry)
380
if text_modified or meta_modified:
384
# TODO 20060511 (mbp, rbc) factor out 'detect_rename' here.
385
if old_entry.parent_id != new_entry.parent_id:
387
elif old_entry.name != new_entry.name:
391
if renamed and not modified:
392
return InventoryEntry.RENAMED
393
if modified and not renamed:
395
if modified and renamed:
396
return InventoryEntry.MODIFIED_AND_RENAMED
400
return ("%s(%r, %r, parent_id=%r, revision=%r)"
401
% (self.__class__.__name__,
407
def snapshot(self, revision, path, previous_entries,
408
work_tree, commit_builder):
409
"""Make a snapshot of this entry which may or may not have changed.
411
This means that all its fields are populated, that it has its
412
text stored in the text store or weave.
414
mutter('new parents of %s are %r', path, previous_entries)
415
self._read_tree_state(path, work_tree)
416
# TODO: Where should we determine whether to reuse a
417
# previous revision id or create a new revision? 20060606
418
if len(previous_entries) == 1:
419
# cannot be unchanged unless there is only one parent file rev.
420
parent_ie = previous_entries.values()[0]
421
if self._unchanged(parent_ie):
422
mutter("found unchanged entry")
423
self.revision = parent_ie.revision
425
return self._snapshot_into_revision(revision, previous_entries,
426
work_tree, commit_builder)
428
def _snapshot_into_revision(self, revision, previous_entries, work_tree,
430
"""Record this revision unconditionally into a store.
432
The entry's last-changed revision property (`revision`) is updated to
433
that of the new revision.
435
:param revision: id of the new revision that is being recorded.
437
:returns: String description of the commit (e.g. "merged", "modified"), etc.
439
mutter('new revision {%s} for {%s}', revision, self.file_id)
440
self.revision = revision
441
self._snapshot_text(previous_entries, work_tree, commit_builder)
443
def _snapshot_text(self, file_parents, work_tree, commit_builder):
444
"""Record the 'text' of this entry, whatever form that takes.
446
This default implementation simply adds an empty text.
448
raise NotImplementedError(self._snapshot_text)
450
def __eq__(self, other):
451
if not isinstance(other, InventoryEntry):
452
return NotImplemented
454
return ((self.file_id == other.file_id)
455
and (self.name == other.name)
456
and (other.symlink_target == self.symlink_target)
457
and (self.text_sha1 == other.text_sha1)
458
and (self.text_size == other.text_size)
459
and (self.text_id == other.text_id)
460
and (self.parent_id == other.parent_id)
461
and (self.kind == other.kind)
462
and (self.revision == other.revision)
463
and (self.executable == other.executable)
466
def __ne__(self, other):
467
return not (self == other)
470
raise ValueError('not hashable')
472
def _unchanged(self, previous_ie):
473
"""Has this entry changed relative to previous_ie.
475
This method should be overridden in child classes.
478
# different inv parent
479
if previous_ie.parent_id != self.parent_id:
482
elif previous_ie.name != self.name:
486
def _read_tree_state(self, path, work_tree):
487
"""Populate fields in the inventory entry from the given tree.
489
Note that this should be modified to be a noop on virtual trees
490
as all entries created there are prepopulated.
492
# TODO: Rather than running this manually, we should check the
493
# working sha1 and other expensive properties when they're
494
# first requested, or preload them if they're already known
495
pass # nothing to do by default
497
def _forget_tree_state(self):
501
class RootEntry(InventoryEntry):
503
def _check(self, checker, rev_id, tree):
504
"""See InventoryEntry._check"""
506
def __init__(self, file_id):
507
self.file_id = file_id
509
self.kind = 'root_directory'
510
self.parent_id = None
514
def __eq__(self, other):
515
if not isinstance(other, RootEntry):
516
return NotImplemented
518
return (self.file_id == other.file_id) \
519
and (self.children == other.children)
522
class InventoryDirectory(InventoryEntry):
523
"""A directory in an inventory."""
525
def _check(self, checker, rev_id, tree):
526
"""See InventoryEntry._check"""
527
if self.text_sha1 != None or self.text_size != None or self.text_id != None:
528
raise BzrCheckError('directory {%s} has text in revision {%s}'
529
% (self.file_id, rev_id))
532
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
533
other.revision = self.revision
534
# note that children are *not* copied; they're pulled across when
538
def __init__(self, file_id, name, parent_id):
539
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
541
self.kind = 'directory'
543
def kind_character(self):
544
"""See InventoryEntry.kind_character."""
547
def _put_in_tar(self, item, tree):
548
"""See InventoryEntry._put_in_tar."""
549
item.type = tarfile.DIRTYPE
556
def _put_on_disk(self, fullpath, tree):
557
"""See InventoryEntry._put_on_disk."""
560
def _snapshot_text(self, file_parents, work_tree, commit_builder):
561
"""See InventoryEntry._snapshot_text."""
562
commit_builder.modified_directory(self.file_id, file_parents)
565
class InventoryFile(InventoryEntry):
566
"""A file in an inventory."""
568
def _check(self, checker, tree_revision_id, tree):
569
"""See InventoryEntry._check"""
570
t = (self.file_id, self.revision)
571
if t in checker.checked_texts:
572
prev_sha = checker.checked_texts[t]
573
if prev_sha != self.text_sha1:
574
raise BzrCheckError('mismatched sha1 on {%s} in {%s}' %
575
(self.file_id, tree_revision_id))
577
checker.repeated_text_cnt += 1
580
if self.file_id not in checker.checked_weaves:
581
mutter('check weave {%s}', self.file_id)
582
w = tree.get_weave(self.file_id)
583
# Not passing a progress bar, because it creates a new
584
# progress, which overwrites the current progress,
585
# and doesn't look nice
587
checker.checked_weaves[self.file_id] = True
589
w = tree.get_weave(self.file_id)
591
mutter('check version {%s} of {%s}', tree_revision_id, self.file_id)
592
checker.checked_text_cnt += 1
593
# We can't check the length, because Weave doesn't store that
594
# information, and the whole point of looking at the weave's
595
# sha1sum is that we don't have to extract the text.
596
if self.text_sha1 != w.get_sha1(self.revision):
597
raise BzrCheckError('text {%s} version {%s} wrong sha1'
598
% (self.file_id, self.revision))
599
checker.checked_texts[t] = self.text_sha1
602
other = InventoryFile(self.file_id, self.name, self.parent_id)
603
other.executable = self.executable
604
other.text_id = self.text_id
605
other.text_sha1 = self.text_sha1
606
other.text_size = self.text_size
607
other.revision = self.revision
610
def detect_changes(self, old_entry):
611
"""See InventoryEntry.detect_changes."""
612
assert self.text_sha1 != None
613
assert old_entry.text_sha1 != None
614
text_modified = (self.text_sha1 != old_entry.text_sha1)
615
meta_modified = (self.executable != old_entry.executable)
616
return text_modified, meta_modified
618
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
619
output_to, reverse=False):
620
"""See InventoryEntry._diff."""
622
from_text = tree.get_file(self.file_id).readlines()
624
to_text = to_tree.get_file(to_entry.file_id).readlines()
628
text_diff(from_label, from_text,
629
to_label, to_text, output_to)
631
text_diff(to_label, to_text,
632
from_label, from_text, output_to)
635
label_pair = (to_label, from_label)
637
label_pair = (from_label, to_label)
638
print >> output_to, "Binary files %s and %s differ" % label_pair
641
"""See InventoryEntry.has_text."""
644
def __init__(self, file_id, name, parent_id):
645
super(InventoryFile, self).__init__(file_id, name, parent_id)
648
def kind_character(self):
649
"""See InventoryEntry.kind_character."""
652
def _put_in_tar(self, item, tree):
653
"""See InventoryEntry._put_in_tar."""
654
item.type = tarfile.REGTYPE
655
fileobj = tree.get_file(self.file_id)
656
item.size = self.text_size
657
if tree.is_executable(self.file_id):
663
def _put_on_disk(self, fullpath, tree):
664
"""See InventoryEntry._put_on_disk."""
665
pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
666
if tree.is_executable(self.file_id):
667
os.chmod(fullpath, 0755)
669
def _read_tree_state(self, path, work_tree):
670
"""See InventoryEntry._read_tree_state."""
671
self.text_sha1 = work_tree.get_file_sha1(self.file_id, path=path)
672
# FIXME: 20050930 probe for the text size when getting sha1
673
# in _read_tree_state
674
self.executable = work_tree.is_executable(self.file_id, path=path)
677
return ("%s(%r, %r, parent_id=%r, sha1=%r, len=%s)"
678
% (self.__class__.__name__,
685
def _forget_tree_state(self):
686
self.text_sha1 = None
687
self.executable = None
689
def _snapshot_text(self, file_parents, work_tree, commit_builder):
690
"""See InventoryEntry._snapshot_text."""
691
def get_content_byte_lines():
692
return work_tree.get_file(self.file_id).readlines()
693
self.text_sha1, self.text_size = commit_builder.modified_file_text(
694
self.file_id, file_parents, get_content_byte_lines, self.text_sha1, self.text_size)
696
def _unchanged(self, previous_ie):
697
"""See InventoryEntry._unchanged."""
698
compatible = super(InventoryFile, self)._unchanged(previous_ie)
699
if self.text_sha1 != previous_ie.text_sha1:
702
# FIXME: 20050930 probe for the text size when getting sha1
703
# in _read_tree_state
704
self.text_size = previous_ie.text_size
705
if self.executable != previous_ie.executable:
710
class InventoryLink(InventoryEntry):
711
"""A file in an inventory."""
713
__slots__ = ['symlink_target']
715
def _check(self, checker, rev_id, tree):
716
"""See InventoryEntry._check"""
717
if self.text_sha1 != None or self.text_size != None or self.text_id != None:
718
raise BzrCheckError('symlink {%s} has text in revision {%s}'
719
% (self.file_id, rev_id))
720
if self.symlink_target is None:
721
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
722
% (self.file_id, rev_id))
725
other = InventoryLink(self.file_id, self.name, self.parent_id)
726
other.symlink_target = self.symlink_target
727
other.revision = self.revision
730
def detect_changes(self, old_entry):
731
"""See InventoryEntry.detect_changes."""
732
# FIXME: which _modified field should we use ? RBC 20051003
733
text_modified = (self.symlink_target != old_entry.symlink_target)
735
mutter(" symlink target changed")
736
meta_modified = False
737
return text_modified, meta_modified
739
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
740
output_to, reverse=False):
741
"""See InventoryEntry._diff."""
742
from_text = self.symlink_target
743
if to_entry is not None:
744
to_text = to_entry.symlink_target
749
print >>output_to, '=== target changed %r => %r' % (from_text, to_text)
752
print >>output_to, '=== target was %r' % self.symlink_target
754
print >>output_to, '=== target is %r' % self.symlink_target
756
def __init__(self, file_id, name, parent_id):
757
super(InventoryLink, self).__init__(file_id, name, parent_id)
758
self.kind = 'symlink'
760
def kind_character(self):
761
"""See InventoryEntry.kind_character."""
764
def _put_in_tar(self, item, tree):
765
"""See InventoryEntry._put_in_tar."""
766
item.type = tarfile.SYMTYPE
770
item.linkname = self.symlink_target
773
def _put_on_disk(self, fullpath, tree):
774
"""See InventoryEntry._put_on_disk."""
776
os.symlink(self.symlink_target, fullpath)
778
raise BzrError("Failed to create symlink %r -> %r, error: %s" % (fullpath, self.symlink_target, e))
780
def _read_tree_state(self, path, work_tree):
781
"""See InventoryEntry._read_tree_state."""
782
self.symlink_target = work_tree.get_symlink_target(self.file_id)
784
def _forget_tree_state(self):
785
self.symlink_target = None
787
def _unchanged(self, previous_ie):
788
"""See InventoryEntry._unchanged."""
789
compatible = super(InventoryLink, self)._unchanged(previous_ie)
790
if self.symlink_target != previous_ie.symlink_target:
794
def _snapshot_text(self, file_parents, work_tree, commit_builder):
795
"""See InventoryEntry._snapshot_text."""
796
commit_builder.modified_link(
797
self.file_id, file_parents, self.symlink_target)
800
class Inventory(object):
801
"""Inventory of versioned files in a tree.
803
This describes which file_id is present at each point in the tree,
804
and possibly the SHA-1 or other information about the file.
805
Entries can be looked up either by path or by file_id.
807
The inventory represents a typical unix file tree, with
808
directories containing files and subdirectories. We never store
809
the full path to a file, because renaming a directory implicitly
810
moves all of its contents. This class internally maintains a
811
lookup tree that allows the children under a directory to be
814
InventoryEntry objects must not be modified after they are
815
inserted, other than through the Inventory API.
817
>>> inv = Inventory()
818
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
819
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
820
>>> inv['123-123'].name
823
May be treated as an iterator or set to look up file ids:
825
>>> bool(inv.path2id('hello.c'))
830
May also look up by name:
832
>>> [x[0] for x in inv.iter_entries()]
834
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
835
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
836
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT-12345678-12345678', sha1=None, len=None)
838
def __init__(self, root_id=ROOT_ID, revision_id=None):
839
"""Create or read an inventory.
841
If a working directory is specified, the inventory is read
842
from there. If the file is specified, read from that. If not,
843
the inventory is created empty.
845
The inventory is created with a default root directory, with
848
# We are letting Branch.create() create a unique inventory
849
# root id. Rather than generating a random one here.
851
# root_id = bzrlib.branch.gen_file_id('TREE_ROOT')
852
self.root = RootEntry(root_id)
853
# FIXME: this isn't ever used, changing it to self.revision may break
854
# things. TODO make everything use self.revision_id
855
self.revision_id = revision_id
856
self._byid = {self.root.file_id: self.root}
859
# TODO: jam 20051218 Should copy also copy the revision_id?
860
other = Inventory(self.root.file_id)
861
# copy recursively so we know directories will be added before
862
# their children. There are more efficient ways than this...
863
for path, entry in self.iter_entries():
864
if entry == self.root:
866
other.add(entry.copy())
870
return iter(self._byid)
873
"""Returns number of entries."""
874
return len(self._byid)
876
def iter_entries(self, from_dir=None):
877
"""Return (path, entry) pairs, in order by name."""
881
elif isinstance(from_dir, basestring):
882
from_dir = self._byid[from_dir]
884
# unrolling the recursive called changed the time from
885
# 440ms/663ms (inline/total) to 116ms/116ms
886
children = from_dir.children.items()
888
children = collections.deque(children)
889
stack = [(u'', children)]
891
from_dir_relpath, children = stack[-1]
894
name, ie = children.popleft()
896
# we know that from_dir_relpath never ends in a slash
897
# and 'f' doesn't begin with one, we can do a string op, rather
898
# than the checks of pathjoin(), though this means that all paths
900
path = from_dir_relpath + '/' + name
904
if ie.kind != 'directory':
907
# But do this child first
908
new_children = ie.children.items()
910
new_children = collections.deque(new_children)
911
stack.append((path, new_children))
912
# Break out of inner loop, so that we start outer loop with child
915
# if we finished all children, pop it off the stack
918
def iter_entries_by_dir(self, from_dir=None):
919
"""Iterate over the entries in a directory first order.
921
This returns all entries for a directory before returning
922
the entries for children of a directory. This is not
923
lexicographically sorted order, and is a hybrid between
924
depth-first and breadth-first.
926
:return: This yields (path, entry) pairs
928
# TODO? Perhaps this should return the from_dir so that the root is
929
# yielded? or maybe an option?
933
elif isinstance(from_dir, basestring):
934
from_dir = self._byid[from_dir]
936
stack = [(u'', from_dir)]
938
cur_relpath, cur_dir = stack.pop()
941
for child_name, child_ie in sorted(cur_dir.children.iteritems()):
943
child_relpath = cur_relpath + child_name
945
yield child_relpath, child_ie
947
if child_ie.kind == 'directory':
948
child_dirs.append((child_relpath+'/', child_ie))
949
stack.extend(reversed(child_dirs))
952
"""Return list of (path, ie) for all entries except the root.
954
This may be faster than iter_entries.
957
def descend(dir_ie, dir_path):
958
kids = dir_ie.children.items()
960
for name, ie in kids:
961
child_path = pathjoin(dir_path, name)
962
accum.append((child_path, ie))
963
if ie.kind == 'directory':
964
descend(ie, child_path)
966
descend(self.root, u'')
969
def directories(self):
970
"""Return (path, entry) pairs for all directories, including the root.
973
def descend(parent_ie, parent_path):
974
accum.append((parent_path, parent_ie))
976
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
979
for name, child_ie in kids:
980
child_path = pathjoin(parent_path, name)
981
descend(child_ie, child_path)
982
descend(self.root, u'')
985
def __contains__(self, file_id):
986
"""True if this entry contains a file with given id.
988
>>> inv = Inventory()
989
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
990
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
996
return file_id in self._byid
998
def __getitem__(self, file_id):
999
"""Return the entry for given file_id.
1001
>>> inv = Inventory()
1002
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
1003
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
1004
>>> inv['123123'].name
1008
return self._byid[file_id]
1011
raise BzrError("can't look up file_id None")
1013
raise BzrError("file_id {%s} not in inventory" % file_id)
1015
def get_file_kind(self, file_id):
1016
return self._byid[file_id].kind
1018
def get_child(self, parent_id, filename):
1019
return self[parent_id].children.get(filename)
1021
def add(self, entry):
1022
"""Add entry to inventory.
1024
To add a file to a branch ready to be committed, use Branch.add,
1027
Returns the new entry object.
1029
if entry.file_id in self._byid:
1030
raise BzrError("inventory already contains entry with id {%s}" % entry.file_id)
1032
if entry.parent_id == ROOT_ID or entry.parent_id is None:
1033
entry.parent_id = self.root.file_id
1036
parent = self._byid[entry.parent_id]
1038
raise BzrError("parent_id {%s} not in inventory" % entry.parent_id)
1040
if parent.children.has_key(entry.name):
1041
raise BzrError("%s is already versioned" %
1042
pathjoin(self.id2path(parent.file_id), entry.name))
1044
self._byid[entry.file_id] = entry
1045
parent.children[entry.name] = entry
1048
def add_path(self, relpath, kind, file_id=None, parent_id=None):
1049
"""Add entry from a path.
1051
The immediate parent must already be versioned.
1053
Returns the new entry object."""
1055
parts = bzrlib.osutils.splitpath(relpath)
1059
file_id = bzrlib.workingtree.gen_root_id()
1060
self.root = RootEntry(file_id)
1061
self._byid = {self.root.file_id: self.root}
1064
parent_path = parts[:-1]
1065
parent_id = self.path2id(parent_path)
1066
if parent_id is None:
1067
raise NotVersionedError(path=parent_path)
1068
ie = make_entry(kind, parts[-1], parent_id, file_id)
1071
def __delitem__(self, file_id):
1072
"""Remove entry by id.
1074
>>> inv = Inventory()
1075
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1076
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
1085
assert ie.parent_id is None or \
1086
self[ie.parent_id].children[ie.name] == ie
1088
del self._byid[file_id]
1089
if ie.parent_id is not None:
1090
del self[ie.parent_id].children[ie.name]
1092
def __eq__(self, other):
1093
"""Compare two sets by comparing their contents.
1095
>>> i1 = Inventory()
1096
>>> i2 = Inventory()
1099
>>> i1.add(InventoryFile('123', 'foo', ROOT_ID))
1100
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
1103
>>> i2.add(InventoryFile('123', 'foo', ROOT_ID))
1104
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
1108
if not isinstance(other, Inventory):
1109
return NotImplemented
1111
return self._byid == other._byid
1113
def __ne__(self, other):
1114
return not self.__eq__(other)
1117
raise ValueError('not hashable')
1119
def _iter_file_id_parents(self, file_id):
1120
"""Yield the parents of file_id up to the root."""
1121
while file_id != None:
1123
ie = self._byid[file_id]
1125
raise BzrError("file_id {%s} not found in inventory" % file_id)
1127
file_id = ie.parent_id
1129
def get_idpath(self, file_id):
1130
"""Return a list of file_ids for the path to an entry.
1132
The list contains one element for each directory followed by
1133
the id of the file itself. So the length of the returned list
1134
is equal to the depth of the file in the tree, counting the
1135
root directory as depth 1.
1138
for parent in self._iter_file_id_parents(file_id):
1139
p.insert(0, parent.file_id)
1142
def id2path(self, file_id):
1143
"""Return as a string the path to file_id.
1146
>>> e = i.add(InventoryDirectory('src-id', 'src', ROOT_ID))
1147
>>> e = i.add(InventoryFile('foo-id', 'foo.c', parent_id='src-id'))
1148
>>> print i.id2path('foo-id')
1151
# get all names, skipping root
1152
return '/'.join(reversed(
1153
[parent.name for parent in
1154
self._iter_file_id_parents(file_id)][:-1]))
1156
def path2id(self, name):
1157
"""Walk down through directories to return entry of last component.
1159
names may be either a list of path components, or a single
1160
string, in which case it is automatically split.
1162
This returns the entry of the last component in the path,
1163
which may be either a file or a directory.
1165
Returns None IFF the path is not found.
1167
if isinstance(name, types.StringTypes):
1168
name = splitpath(name)
1170
# mutter("lookup path %r" % name)
1175
cie = parent.children[f]
1176
assert cie.name == f
1177
assert cie.parent_id == parent.file_id
1180
# or raise an error?
1183
return parent.file_id
1185
def has_filename(self, names):
1186
return bool(self.path2id(names))
1188
def has_id(self, file_id):
1189
return self._byid.has_key(file_id)
1191
def rename(self, file_id, new_parent_id, new_name):
1192
"""Move a file within the inventory.
1194
This can change either the name, or the parent, or both.
1196
This does not move the working file."""
1197
if not is_valid_name(new_name):
1198
raise BzrError("not an acceptable filename: %r" % new_name)
1200
new_parent = self._byid[new_parent_id]
1201
if new_name in new_parent.children:
1202
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
1204
new_parent_idpath = self.get_idpath(new_parent_id)
1205
if file_id in new_parent_idpath:
1206
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
1207
% (self.id2path(file_id), self.id2path(new_parent_id)))
1209
file_ie = self._byid[file_id]
1210
old_parent = self._byid[file_ie.parent_id]
1212
# TODO: Don't leave things messed up if this fails
1214
del old_parent.children[file_ie.name]
1215
new_parent.children[new_name] = file_ie
1217
file_ie.name = new_name
1218
file_ie.parent_id = new_parent_id
1221
def make_entry(kind, name, parent_id, file_id=None):
1222
"""Create an inventory entry.
1224
:param kind: the type of inventory entry to create.
1225
:param name: the basename of the entry.
1226
:param parent_id: the parent_id of the entry.
1227
:param file_id: the file_id to use. if None, one will be created.
1230
file_id = bzrlib.workingtree.gen_file_id(name)
1231
if kind == 'directory':
1232
return InventoryDirectory(file_id, name, parent_id)
1233
elif kind == 'file':
1234
return InventoryFile(file_id, name, parent_id)
1235
elif kind == 'symlink':
1236
return InventoryLink(file_id, name, parent_id)
1238
raise BzrError("unknown kind %r" % kind)
1244
def is_valid_name(name):
1246
if _NAME_RE is None:
1247
_NAME_RE = re.compile(r'^[^/\\]+$')
1249
return bool(_NAME_RE.match(name))
added
removed
bzrlib/inventory.py
