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
# Constants returned by describe_change()
119
# TODO: These should probably move to some kind of FileChangeDescription
120
# class; that's like what's inside a TreeDelta but we want to be able to
121
# generate them just for one file at a time.
123
MODIFIED_AND_RENAMED = 'modified and renamed'
125
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
126
'text_id', 'parent_id', 'children', 'executable',
129
def _add_text_to_weave(self, new_lines, parents, weave_store, transaction):
130
versionedfile = weave_store.get_weave_or_empty(self.file_id,
132
versionedfile.add_lines(self.revision, parents, new_lines)
133
versionedfile.clear_cache()
135
def detect_changes(self, old_entry):
136
"""Return a (text_modified, meta_modified) from this to old_entry.
138
_read_tree_state must have been called on self and old_entry prior to
139
calling detect_changes.
143
def diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
144
output_to, reverse=False):
145
"""Perform a diff from this to to_entry.
147
text_diff will be used for textual difference calculation.
148
This is a template method, override _diff in child classes.
150
self._read_tree_state(tree.id2path(self.file_id), tree)
152
# cannot diff from one kind to another - you must do a removal
153
# and an addif they do not match.
154
assert self.kind == to_entry.kind
155
to_entry._read_tree_state(to_tree.id2path(to_entry.file_id),
157
self._diff(text_diff, from_label, tree, to_label, to_entry, to_tree,
160
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
161
output_to, reverse=False):
162
"""Perform a diff between two entries of the same kind."""
164
def find_previous_heads(self, previous_inventories,
165
versioned_file_store,
168
"""Return the revisions and entries that directly preceed this.
170
Returned as a map from revision to inventory entry.
172
This is a map containing the file revisions in all parents
173
for which the file exists, and its revision is not a parent of
174
any other. If the file is new, the set will be empty.
176
:param versioned_file_store: A store where ancestry data on this
177
file id can be queried.
178
:param transaction: The transaction that queries to the versioned
179
file store should be completed under.
180
:param entry_vf: The entry versioned file, if its already available.
182
def get_ancestors(weave, entry):
183
return set(weave.get_ancestry(entry.revision))
184
# revision:ie mapping for each ie found in previous_inventories.
186
# revision:ie mapping with one revision for each head.
188
# revision: ancestor list for each head
190
# identify candidate head revision ids.
191
for inv in previous_inventories:
192
if self.file_id in inv:
193
ie = inv[self.file_id]
194
assert ie.file_id == self.file_id
195
if ie.revision in candidates:
196
# same revision value in two different inventories:
197
# correct possible inconsistencies:
198
# * there was a bug in revision updates with 'x' bit
201
if candidates[ie.revision].executable != ie.executable:
202
candidates[ie.revision].executable = False
203
ie.executable = False
204
except AttributeError:
206
# must now be the same.
207
assert candidates[ie.revision] == ie
209
# add this revision as a candidate.
210
candidates[ie.revision] = ie
212
# common case optimisation
213
if len(candidates) == 1:
214
# if there is only one candidate revision found
215
# then we can opening the versioned file to access ancestry:
216
# there cannot be any ancestors to eliminate when there is
217
# only one revision available.
218
heads[ie.revision] = ie
221
# eliminate ancestors amongst the available candidates:
222
# heads are those that are not an ancestor of any other candidate
223
# - this provides convergence at a per-file level.
224
for ie in candidates.values():
225
# may be an ancestor of a known head:
226
already_present = 0 != len(
227
[head for head in heads
228
if ie.revision in head_ancestors[head]])
230
# an ancestor of an analyzed candidate.
232
# not an ancestor of a known head:
233
# load the versioned file for this file id if needed
235
entry_vf = versioned_file_store.get_weave_or_empty(
236
self.file_id, transaction)
237
ancestors = get_ancestors(entry_vf, ie)
238
# may knock something else out:
239
check_heads = list(heads.keys())
240
for head in check_heads:
241
if head in ancestors:
242
# this previously discovered 'head' is not
243
# really a head - its an ancestor of the newly
246
head_ancestors[ie.revision] = ancestors
247
heads[ie.revision] = ie
250
def get_tar_item(self, root, dp, now, tree):
251
"""Get a tarfile item and a file stream for its content."""
252
item = tarfile.TarInfo(pathjoin(root, dp))
253
# TODO: would be cool to actually set it to the timestamp of the
254
# revision it was last changed
256
fileobj = self._put_in_tar(item, tree)
260
"""Return true if the object this entry represents has textual data.
262
Note that textual data includes binary content.
264
Also note that all entries get weave files created for them.
265
This attribute is primarily used when upgrading from old trees that
266
did not have the weave index for all inventory entries.
270
def __init__(self, file_id, name, parent_id, text_id=None):
271
"""Create an InventoryEntry
273
The filename must be a single component, relative to the
274
parent directory; it cannot be a whole path or relative name.
276
>>> e = InventoryFile('123', 'hello.c', ROOT_ID)
281
>>> e = InventoryFile('123', 'src/hello.c', ROOT_ID)
282
Traceback (most recent call last):
283
InvalidEntryName: Invalid entry name: src/hello.c
285
assert isinstance(name, basestring), name
286
if '/' in name or '\\' in name:
287
raise InvalidEntryName(name=name)
288
self.executable = False
290
self.text_sha1 = None
291
self.text_size = None
292
self.file_id = file_id
294
self.text_id = text_id
295
self.parent_id = parent_id
296
self.symlink_target = None
298
def kind_character(self):
299
"""Return a short kind indicator useful for appending to names."""
300
raise BzrError('unknown kind %r' % self.kind)
302
known_kinds = ('file', 'directory', 'symlink', 'root_directory')
304
def _put_in_tar(self, item, tree):
305
"""populate item for stashing in a tar, and return the content stream.
307
If no content is available, return None.
309
raise BzrError("don't know how to export {%s} of kind %r" %
310
(self.file_id, self.kind))
312
def put_on_disk(self, dest, dp, tree):
313
"""Create a representation of self on disk in the prefix dest.
315
This is a template method - implement _put_on_disk in subclasses.
317
fullpath = pathjoin(dest, dp)
318
self._put_on_disk(fullpath, tree)
319
mutter(" export {%s} kind %s to %s", self.file_id,
322
def _put_on_disk(self, fullpath, tree):
323
"""Put this entry onto disk at fullpath, from tree tree."""
324
raise BzrError("don't know how to export {%s} of kind %r" % (self.file_id, self.kind))
326
def sorted_children(self):
327
l = self.children.items()
332
def versionable_kind(kind):
333
return kind in ('file', 'directory', 'symlink')
335
def check(self, checker, rev_id, inv, tree):
336
"""Check this inventory entry is intact.
338
This is a template method, override _check for kind specific
341
:param checker: Check object providing context for the checks;
342
can be used to find out what parts of the repository have already
344
:param rev_id: Revision id from which this InventoryEntry was loaded.
345
Not necessarily the last-changed revision for this file.
346
:param inv: Inventory from which the entry was loaded.
347
:param tree: RevisionTree for this entry.
349
if self.parent_id != None:
350
if not inv.has_id(self.parent_id):
351
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
352
% (self.parent_id, rev_id))
353
self._check(checker, rev_id, tree)
355
def _check(self, checker, rev_id, tree):
356
"""Check this inventory entry for kind specific errors."""
357
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
361
"""Clone this inventory entry."""
362
raise NotImplementedError
365
def describe_change(old_entry, new_entry):
366
"""Describe the change between old_entry and this.
368
This smells of being an InterInventoryEntry situation, but as its
369
the first one, we're making it a static method for now.
371
An entry with a different parent, or different name is considered
372
to be renamed. Reparenting is an internal detail.
373
Note that renaming the parent does not trigger a rename for the
376
# TODO: Perhaps return an object rather than just a string
377
if old_entry is new_entry:
378
# also the case of both being None
380
elif old_entry is None:
382
elif new_entry is None:
384
text_modified, meta_modified = new_entry.detect_changes(old_entry)
385
if text_modified or meta_modified:
389
# TODO 20060511 (mbp, rbc) factor out 'detect_rename' here.
390
if old_entry.parent_id != new_entry.parent_id:
392
elif old_entry.name != new_entry.name:
396
if renamed and not modified:
397
return InventoryEntry.RENAMED
398
if modified and not renamed:
400
if modified and renamed:
401
return InventoryEntry.MODIFIED_AND_RENAMED
405
return ("%s(%r, %r, parent_id=%r)"
406
% (self.__class__.__name__,
411
def snapshot(self, revision, path, previous_entries,
412
work_tree, weave_store, transaction):
413
"""Make a snapshot of this entry which may or may not have changed.
415
This means that all its fields are populated, that it has its
416
text stored in the text store or weave.
418
mutter('new parents of %s are %r', path, previous_entries)
419
self._read_tree_state(path, work_tree)
420
if len(previous_entries) == 1:
421
# cannot be unchanged unless there is only one parent file rev.
422
parent_ie = previous_entries.values()[0]
423
if self._unchanged(parent_ie):
424
mutter("found unchanged entry")
425
self.revision = parent_ie.revision
427
return self._snapshot_into_revision(revision, previous_entries,
428
work_tree, weave_store, transaction)
430
def _snapshot_into_revision(self, revision, previous_entries, work_tree,
431
weave_store, transaction):
432
"""Record this revision unconditionally into a store.
434
The entry's last-changed revision property (`revision`) is updated to
435
that of the new revision.
437
:param revision: id of the new revision that is being recorded.
439
:returns: String description of the commit (e.g. "merged", "modified"), etc.
441
mutter('new revision {%s} for {%s}', revision, self.file_id)
442
self.revision = revision
443
self._snapshot_text(previous_entries, work_tree, weave_store,
446
def _snapshot_text(self, file_parents, work_tree, weave_store, transaction):
447
"""Record the 'text' of this entry, whatever form that takes.
449
This default implementation simply adds an empty text.
451
mutter('storing file {%s} in revision {%s}',
452
self.file_id, self.revision)
453
self._add_text_to_weave([], file_parents.keys(), weave_store, transaction)
455
def __eq__(self, other):
456
if not isinstance(other, InventoryEntry):
457
return NotImplemented
459
return ((self.file_id == other.file_id)
460
and (self.name == other.name)
461
and (other.symlink_target == self.symlink_target)
462
and (self.text_sha1 == other.text_sha1)
463
and (self.text_size == other.text_size)
464
and (self.text_id == other.text_id)
465
and (self.parent_id == other.parent_id)
466
and (self.kind == other.kind)
467
and (self.revision == other.revision)
468
and (self.executable == other.executable)
471
def __ne__(self, other):
472
return not (self == other)
475
raise ValueError('not hashable')
477
def _unchanged(self, previous_ie):
478
"""Has this entry changed relative to previous_ie.
480
This method should be overriden in child classes.
483
# different inv parent
484
if previous_ie.parent_id != self.parent_id:
487
elif previous_ie.name != self.name:
491
def _read_tree_state(self, path, work_tree):
492
"""Populate fields in the inventory entry from the given tree.
494
Note that this should be modified to be a noop on virtual trees
495
as all entries created there are prepopulated.
497
# TODO: Rather than running this manually, we should check the
498
# working sha1 and other expensive properties when they're
499
# first requested, or preload them if they're already known
500
pass # nothing to do by default
502
def _forget_tree_state(self):
506
class RootEntry(InventoryEntry):
508
def _check(self, checker, rev_id, tree):
509
"""See InventoryEntry._check"""
511
def __init__(self, file_id):
512
self.file_id = file_id
514
self.kind = 'root_directory'
515
self.parent_id = None
518
def __eq__(self, other):
519
if not isinstance(other, RootEntry):
520
return NotImplemented
522
return (self.file_id == other.file_id) \
523
and (self.children == other.children)
526
class InventoryDirectory(InventoryEntry):
527
"""A directory in an inventory."""
529
def _check(self, checker, rev_id, tree):
530
"""See InventoryEntry._check"""
531
if self.text_sha1 != None or self.text_size != None or self.text_id != None:
532
raise BzrCheckError('directory {%s} has text in revision {%s}'
533
% (self.file_id, rev_id))
536
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
537
other.revision = self.revision
538
# note that children are *not* copied; they're pulled across when
542
def __init__(self, file_id, name, parent_id):
543
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
545
self.kind = 'directory'
547
def kind_character(self):
548
"""See InventoryEntry.kind_character."""
551
def _put_in_tar(self, item, tree):
552
"""See InventoryEntry._put_in_tar."""
553
item.type = tarfile.DIRTYPE
560
def _put_on_disk(self, fullpath, tree):
561
"""See InventoryEntry._put_on_disk."""
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)
672
self.executable = work_tree.is_executable(self.file_id)
674
def _forget_tree_state(self):
675
self.text_sha1 = None
676
self.executable = None
678
def _snapshot_text(self, file_parents, work_tree, versionedfile_store, transaction):
679
"""See InventoryEntry._snapshot_text."""
680
mutter('storing text of file {%s} in revision {%s} into %r',
681
self.file_id, self.revision, versionedfile_store)
682
# special case to avoid diffing on renames or
684
if (len(file_parents) == 1
685
and self.text_sha1 == file_parents.values()[0].text_sha1
686
and self.text_size == file_parents.values()[0].text_size):
687
previous_ie = file_parents.values()[0]
688
versionedfile = versionedfile_store.get_weave(self.file_id, transaction)
689
versionedfile.clone_text(self.revision, previous_ie.revision, file_parents.keys())
691
new_lines = work_tree.get_file(self.file_id).readlines()
692
self._add_text_to_weave(new_lines, file_parents.keys(), versionedfile_store,
694
self.text_sha1 = sha_strings(new_lines)
695
self.text_size = sum(map(len, new_lines))
698
def _unchanged(self, previous_ie):
699
"""See InventoryEntry._unchanged."""
700
compatible = super(InventoryFile, self)._unchanged(previous_ie)
701
if self.text_sha1 != previous_ie.text_sha1:
704
# FIXME: 20050930 probe for the text size when getting sha1
705
# in _read_tree_state
706
self.text_size = previous_ie.text_size
707
if self.executable != previous_ie.executable:
712
class InventoryLink(InventoryEntry):
713
"""A file in an inventory."""
715
__slots__ = ['symlink_target']
717
def _check(self, checker, rev_id, tree):
718
"""See InventoryEntry._check"""
719
if self.text_sha1 != None or self.text_size != None or self.text_id != None:
720
raise BzrCheckError('symlink {%s} has text in revision {%s}'
721
% (self.file_id, rev_id))
722
if self.symlink_target == None:
723
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
724
% (self.file_id, rev_id))
727
other = InventoryLink(self.file_id, self.name, self.parent_id)
728
other.symlink_target = self.symlink_target
729
other.revision = self.revision
732
def detect_changes(self, old_entry):
733
"""See InventoryEntry.detect_changes."""
734
# FIXME: which _modified field should we use ? RBC 20051003
735
text_modified = (self.symlink_target != old_entry.symlink_target)
737
mutter(" symlink target changed")
738
meta_modified = False
739
return text_modified, meta_modified
741
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
742
output_to, reverse=False):
743
"""See InventoryEntry._diff."""
744
from_text = self.symlink_target
745
if to_entry is not None:
746
to_text = to_entry.symlink_target
751
print >>output_to, '=== target changed %r => %r' % (from_text, to_text)
754
print >>output_to, '=== target was %r' % self.symlink_target
756
print >>output_to, '=== target is %r' % self.symlink_target
758
def __init__(self, file_id, name, parent_id):
759
super(InventoryLink, self).__init__(file_id, name, parent_id)
760
self.kind = 'symlink'
762
def kind_character(self):
763
"""See InventoryEntry.kind_character."""
766
def _put_in_tar(self, item, tree):
767
"""See InventoryEntry._put_in_tar."""
768
item.type = tarfile.SYMTYPE
772
item.linkname = self.symlink_target
775
def _put_on_disk(self, fullpath, tree):
776
"""See InventoryEntry._put_on_disk."""
778
os.symlink(self.symlink_target, fullpath)
780
raise BzrError("Failed to create symlink %r -> %r, error: %s" % (fullpath, self.symlink_target, e))
782
def _read_tree_state(self, path, work_tree):
783
"""See InventoryEntry._read_tree_state."""
784
self.symlink_target = work_tree.get_symlink_target(self.file_id)
786
def _forget_tree_state(self):
787
self.symlink_target = None
789
def _unchanged(self, previous_ie):
790
"""See InventoryEntry._unchanged."""
791
compatible = super(InventoryLink, self)._unchanged(previous_ie)
792
if self.symlink_target != previous_ie.symlink_target:
797
class Inventory(object):
798
"""Inventory of versioned files in a tree.
800
This describes which file_id is present at each point in the tree,
801
and possibly the SHA-1 or other information about the file.
802
Entries can be looked up either by path or by file_id.
804
The inventory represents a typical unix file tree, with
805
directories containing files and subdirectories. We never store
806
the full path to a file, because renaming a directory implicitly
807
moves all of its contents. This class internally maintains a
808
lookup tree that allows the children under a directory to be
811
InventoryEntry objects must not be modified after they are
812
inserted, other than through the Inventory API.
814
>>> inv = Inventory()
815
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
816
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT')
817
>>> inv['123-123'].name
820
May be treated as an iterator or set to look up file ids:
822
>>> bool(inv.path2id('hello.c'))
827
May also look up by name:
829
>>> [x[0] for x in inv.iter_entries()]
831
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
832
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
833
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT-12345678-12345678')
835
def __init__(self, root_id=ROOT_ID, revision_id=None):
836
"""Create or read an inventory.
838
If a working directory is specified, the inventory is read
839
from there. If the file is specified, read from that. If not,
840
the inventory is created empty.
842
The inventory is created with a default root directory, with
845
# We are letting Branch.create() create a unique inventory
846
# root id. Rather than generating a random one here.
848
# root_id = bzrlib.branch.gen_file_id('TREE_ROOT')
849
self.root = RootEntry(root_id)
850
self.revision_id = revision_id
851
self._byid = {self.root.file_id: self.root}
855
# TODO: jam 20051218 Should copy also copy the revision_id?
856
other = Inventory(self.root.file_id)
857
# copy recursively so we know directories will be added before
858
# their children. There are more efficient ways than this...
859
for path, entry in self.iter_entries():
860
if entry == self.root:
862
other.add(entry.copy())
867
return iter(self._byid)
871
"""Returns number of entries."""
872
return len(self._byid)
875
def iter_entries(self, from_dir=None):
876
"""Return (path, entry) pairs, in order by name."""
880
elif isinstance(from_dir, basestring):
881
from_dir = self._byid[from_dir]
883
kids = from_dir.children.items()
885
for name, ie in kids:
887
if ie.kind == 'directory':
888
for cn, cie in self.iter_entries(from_dir=ie.file_id):
889
yield pathjoin(name, cn), cie
893
"""Return list of (path, ie) for all entries except the root.
895
This may be faster than iter_entries.
898
def descend(dir_ie, dir_path):
899
kids = dir_ie.children.items()
901
for name, ie in kids:
902
child_path = pathjoin(dir_path, name)
903
accum.append((child_path, ie))
904
if ie.kind == 'directory':
905
descend(ie, child_path)
907
descend(self.root, u'')
911
def directories(self):
912
"""Return (path, entry) pairs for all directories, including the root.
915
def descend(parent_ie, parent_path):
916
accum.append((parent_path, parent_ie))
918
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
921
for name, child_ie in kids:
922
child_path = pathjoin(parent_path, name)
923
descend(child_ie, child_path)
924
descend(self.root, u'')
929
def __contains__(self, file_id):
930
"""True if this entry contains a file with given id.
932
>>> inv = Inventory()
933
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
934
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT')
940
return file_id in self._byid
943
def __getitem__(self, file_id):
944
"""Return the entry for given file_id.
946
>>> inv = Inventory()
947
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
948
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT')
949
>>> inv['123123'].name
953
return self._byid[file_id]
956
raise BzrError("can't look up file_id None")
958
raise BzrError("file_id {%s} not in inventory" % file_id)
961
def get_file_kind(self, file_id):
962
return self._byid[file_id].kind
964
def get_child(self, parent_id, filename):
965
return self[parent_id].children.get(filename)
968
def add(self, entry):
969
"""Add entry to inventory.
971
To add a file to a branch ready to be committed, use Branch.add,
974
Returns the new entry object.
976
if entry.file_id in self._byid:
977
raise BzrError("inventory already contains entry with id {%s}" % entry.file_id)
979
if entry.parent_id == ROOT_ID or entry.parent_id is None:
980
entry.parent_id = self.root.file_id
983
parent = self._byid[entry.parent_id]
985
raise BzrError("parent_id {%s} not in inventory" % entry.parent_id)
987
if parent.children.has_key(entry.name):
988
raise BzrError("%s is already versioned" %
989
pathjoin(self.id2path(parent.file_id), entry.name))
991
self._byid[entry.file_id] = entry
992
parent.children[entry.name] = entry
996
def add_path(self, relpath, kind, file_id=None):
997
"""Add entry from a path.
999
The immediate parent must already be versioned.
1001
Returns the new entry object."""
1002
from bzrlib.workingtree import gen_file_id
1004
parts = bzrlib.osutils.splitpath(relpath)
1007
file_id = gen_file_id(relpath)
1010
self.root = RootEntry(file_id)
1011
self._byid = {self.root.file_id: self.root}
1014
parent_path = parts[:-1]
1015
parent_id = self.path2id(parent_path)
1016
if parent_id == None:
1017
raise NotVersionedError(path=parent_path)
1018
if kind == 'directory':
1019
ie = InventoryDirectory(file_id, parts[-1], parent_id)
1020
elif kind == 'file':
1021
ie = InventoryFile(file_id, parts[-1], parent_id)
1022
elif kind == 'symlink':
1023
ie = InventoryLink(file_id, parts[-1], parent_id)
1025
raise BzrError("unknown kind %r" % kind)
1029
def __delitem__(self, file_id):
1030
"""Remove entry by id.
1032
>>> inv = Inventory()
1033
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1034
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT')
1043
assert ie.parent_id is None or \
1044
self[ie.parent_id].children[ie.name] == ie
1046
del self._byid[file_id]
1047
if ie.parent_id is not None:
1048
del self[ie.parent_id].children[ie.name]
1051
def __eq__(self, other):
1052
"""Compare two sets by comparing their contents.
1054
>>> i1 = Inventory()
1055
>>> i2 = Inventory()
1058
>>> i1.add(InventoryFile('123', 'foo', ROOT_ID))
1059
InventoryFile('123', 'foo', parent_id='TREE_ROOT')
1062
>>> i2.add(InventoryFile('123', 'foo', ROOT_ID))
1063
InventoryFile('123', 'foo', parent_id='TREE_ROOT')
1067
if not isinstance(other, Inventory):
1068
return NotImplemented
1070
if len(self._byid) != len(other._byid):
1071
# shortcut: obviously not the same
1074
return self._byid == other._byid
1077
def __ne__(self, other):
1078
return not self.__eq__(other)
1082
raise ValueError('not hashable')
1084
def _iter_file_id_parents(self, file_id):
1085
"""Yield the parents of file_id up to the root."""
1086
while file_id != None:
1088
ie = self._byid[file_id]
1090
raise BzrError("file_id {%s} not found in inventory" % file_id)
1092
file_id = ie.parent_id
1094
def get_idpath(self, file_id):
1095
"""Return a list of file_ids for the path to an entry.
1097
The list contains one element for each directory followed by
1098
the id of the file itself. So the length of the returned list
1099
is equal to the depth of the file in the tree, counting the
1100
root directory as depth 1.
1103
for parent in self._iter_file_id_parents(file_id):
1104
p.insert(0, parent.file_id)
1107
def id2path(self, file_id):
1108
"""Return as a string the path to file_id.
1111
>>> e = i.add(InventoryDirectory('src-id', 'src', ROOT_ID))
1112
>>> e = i.add(InventoryFile('foo-id', 'foo.c', parent_id='src-id'))
1113
>>> print i.id2path('foo-id')
1116
# get all names, skipping root
1117
return '/'.join(reversed(
1118
[parent.name for parent in
1119
self._iter_file_id_parents(file_id)][:-1]))
1121
def path2id(self, name):
1122
"""Walk down through directories to return entry of last component.
1124
names may be either a list of path components, or a single
1125
string, in which case it is automatically split.
1127
This returns the entry of the last component in the path,
1128
which may be either a file or a directory.
1130
Returns None iff the path is not found.
1132
if isinstance(name, types.StringTypes):
1133
name = splitpath(name)
1135
mutter("lookup path %r" % name)
1140
cie = parent.children[f]
1141
assert cie.name == f
1142
assert cie.parent_id == parent.file_id
1145
# or raise an error?
1148
return parent.file_id
1151
def has_filename(self, names):
1152
return bool(self.path2id(names))
1155
def has_id(self, file_id):
1156
return self._byid.has_key(file_id)
1159
def rename(self, file_id, new_parent_id, new_name):
1160
"""Move a file within the inventory.
1162
This can change either the name, or the parent, or both.
1164
This does not move the working file."""
1165
if not is_valid_name(new_name):
1166
raise BzrError("not an acceptable filename: %r" % new_name)
1168
new_parent = self._byid[new_parent_id]
1169
if new_name in new_parent.children:
1170
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
1172
new_parent_idpath = self.get_idpath(new_parent_id)
1173
if file_id in new_parent_idpath:
1174
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
1175
% (self.id2path(file_id), self.id2path(new_parent_id)))
1177
file_ie = self._byid[file_id]
1178
old_parent = self._byid[file_ie.parent_id]
1180
# TODO: Don't leave things messed up if this fails
1182
del old_parent.children[file_ie.name]
1183
new_parent.children[new_name] = file_ie
1185
file_ie.name = new_name
1186
file_ie.parent_id = new_parent_id
1193
def is_valid_name(name):
1195
if _NAME_RE == None:
1196
_NAME_RE = re.compile(r'^[^/\\]+$')
1198
return bool(_NAME_RE.match(name))
added
removed
bzrlib/inventory.py
