1
# Copyright (C) 2005, 2006, 2007 Canonical Ltd
2
# -*- coding: UTF-8 -*-
3
4
# This program is free software; you can redistribute it and/or modify
4
5
# it under the terms of the GNU General Public License as published by
5
6
# the Free Software Foundation; either version 2 of the License, or
6
7
# (at your option) any later version.
8
9
# This program is distributed in the hope that it will be useful,
9
10
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
11
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
12
# GNU General Public License for more details.
13
14
# You should have received a copy of the GNU General Public License
14
15
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 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.
30
from copy import deepcopy
32
from bzrlib.lazy_import import lazy_import
33
lazy_import(globals(), """
16
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
18
"""Inventories map files to their name in a revision."""
21
__copyright__ = "Copyright (C) 2005 Canonical Ltd."
22
__author__ = "Martin Pool <mbp@canonical.com>"
24
import sys, os.path, types
28
from cElementTree import Element, ElementTree, SubElement
30
from elementtree.ElementTree import Element, ElementTree, SubElement
32
from xml import XMLMixin
33
from errors import bailout
50
from bzrlib.errors import (
54
from bzrlib.symbol_versioning import deprecated_in, deprecated_method
36
from bzrlib.osutils import uuid, quotefn, splitpath, joinpath, appendpath
55
37
from bzrlib.trace import mutter
58
class InventoryEntry(object):
39
class InventoryEntry(XMLMixin):
59
40
"""Description of a versioned file.
61
42
An InventoryEntry has the following fields, which are also
62
43
present in the XML inventory-entry element:
67
(within the parent directory)
70
file_id of the parent directory, or ROOT_ID
73
the revision_id in which this variation of this file was
77
Indicates that this file should be executable on systems
81
sha-1 of the text of the file
84
size in bytes of the text of the file
86
(reading a version 4 tree created a text_id field.)
46
* *name*: (only the basename within the directory, must not
48
* *kind*: "directory" or "file"
49
* *directory_id*: (if absent/null means the branch root directory)
50
* *text_sha1*: only for files
51
* *text_size*: in bytes, only for files
52
* *text_id*: identifier for the text version, only for files
54
InventoryEntries can also exist inside a WorkingTree
55
inventory, in which case they are not yet bound to a
56
particular revision of the file. In that case the text_sha1,
57
text_size and text_id are absent.
88
60
>>> i = Inventory()
91
>>> i.add(InventoryDirectory('123', 'src', ROOT_ID))
92
InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None)
93
>>> i.add(InventoryFile('2323', 'hello.c', parent_id='123'))
94
InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None, revision=None)
95
>>> shouldbe = {0: '', 1: 'src', 2: 'src/hello.c'}
96
>>> for ix, j in enumerate(i.iter_entries()):
97
... print (j[0] == shouldbe[ix], j[1])
62
>>> i.add(InventoryEntry('123', 'src', kind='directory'))
63
>>> i.add(InventoryEntry('2323', 'hello.c', parent_id='123'))
64
>>> for j in i.iter_entries():
67
('src', InventoryEntry('123', 'src', kind='directory', parent_id=None))
68
('src/hello.c', InventoryEntry('2323', 'hello.c', kind='file', parent_id='123'))
69
>>> i.add(InventoryEntry('2323', 'bye.c', parent_id='123'))
70
Traceback (most recent call last):
99
(True, InventoryDirectory('TREE_ROOT', u'', parent_id=None, revision=None))
100
(True, InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None))
101
(True, InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None, revision=None))
102
>>> i.add(InventoryFile('2324', 'bye.c', '123'))
103
InventoryFile('2324', 'bye.c', parent_id='123', sha1=None, len=None, revision=None)
104
>>> i.add(InventoryDirectory('2325', 'wibble', '123'))
105
InventoryDirectory('2325', 'wibble', parent_id='123', revision=None)
72
BzrError: ('inventory already contains entry with id {2323}', [])
73
>>> i.add(InventoryEntry('2324', 'bye.c', parent_id='123'))
74
>>> i.add(InventoryEntry('2325', 'wibble', parent_id='123', kind='directory'))
106
75
>>> i.path2id('src/wibble')
110
>>> i.add(InventoryFile('2326', 'wibble.c', '2325'))
111
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None, revision=None)
79
>>> i.add(InventoryEntry('2326', 'wibble.c', parent_id='2325'))
113
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None, revision=None)
114
>>> for path, entry in i.iter_entries():
81
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
82
>>> for j in i.iter_entries():
84
... assert i.path2id(j[0])
122
90
src/wibble/wibble.c
123
91
>>> i.id2path('2326')
124
92
'src/wibble/wibble.c'
94
:todo: Maybe also keep the full path of the entry, and the children?
95
But those depend on its position within a particular inventory, and
96
it would be nice not to need to hold the backpointer here.
127
# Constants returned by describe_change()
129
# TODO: These should probably move to some kind of FileChangeDescription
130
# class; that's like what's inside a TreeDelta but we want to be able to
131
# generate them just for one file at a time.
133
MODIFIED_AND_RENAMED = 'modified and renamed'
137
def detect_changes(self, old_entry):
138
"""Return a (text_modified, meta_modified) from this to old_entry.
140
_read_tree_state must have been called on self and old_entry prior to
141
calling detect_changes.
145
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
146
output_to, reverse=False):
147
"""Perform a diff between two entries of the same kind."""
149
def parent_candidates(self, previous_inventories):
150
"""Find possible per-file graph parents.
152
This is currently defined by:
153
- Select the last changed revision in the parent inventory.
154
- Do deal with a short lived bug in bzr 0.8's development two entries
155
that have the same last changed but different 'x' bit settings are
158
# revision:ie mapping for each ie found in previous_inventories.
160
# identify candidate head revision ids.
161
for inv in previous_inventories:
162
if self.file_id in inv:
163
ie = inv[self.file_id]
164
if ie.revision in candidates:
165
# same revision value in two different inventories:
166
# correct possible inconsistencies:
167
# * there was a bug in revision updates with 'x' bit
170
if candidates[ie.revision].executable != ie.executable:
171
candidates[ie.revision].executable = False
172
ie.executable = False
173
except AttributeError:
176
# add this revision as a candidate.
177
candidates[ie.revision] = ie
180
@deprecated_method(deprecated_in((1, 6, 0)))
181
def get_tar_item(self, root, dp, now, tree):
182
"""Get a tarfile item and a file stream for its content."""
183
item = tarfile.TarInfo(osutils.pathjoin(root, dp).encode('utf8'))
184
# TODO: would be cool to actually set it to the timestamp of the
185
# revision it was last changed
187
fileobj = self._put_in_tar(item, tree)
191
"""Return true if the object this entry represents has textual data.
193
Note that textual data includes binary content.
195
Also note that all entries get weave files created for them.
196
This attribute is primarily used when upgrading from old trees that
197
did not have the weave index for all inventory entries.
201
def __init__(self, file_id, name, parent_id, text_id=None):
98
def __init__(self, file_id, name, kind='file', text_id=None,
202
100
"""Create an InventoryEntry
204
102
The filename must be a single component, relative to the
205
103
parent directory; it cannot be a whole path or relative name.
207
>>> e = InventoryFile('123', 'hello.c', ROOT_ID)
105
>>> e = InventoryEntry('123', 'hello.c')
212
>>> e = InventoryFile('123', 'src/hello.c', ROOT_ID)
110
>>> e = InventoryEntry('123', 'src/hello.c')
213
111
Traceback (most recent call last):
214
InvalidEntryName: Invalid entry name: src/hello.c
112
BzrError: ("InventoryEntry name is not a simple filename: 'src/hello.c'", [])
216
if '/' in name or '\\' in name:
217
raise errors.InvalidEntryName(name=name)
218
self.executable = False
220
self.text_sha1 = None
221
self.text_size = None
115
if len(splitpath(name)) != 1:
116
bailout('InventoryEntry name is not a simple filename: %r'
222
119
self.file_id = file_id
121
assert kind in ['file', 'directory']
224
123
self.text_id = text_id
225
124
self.parent_id = parent_id
226
self.symlink_target = None
227
self.reference_revision = None
229
def kind_character(self):
230
"""Return a short kind indicator useful for appending to names."""
231
raise BzrError('unknown kind %r' % self.kind)
233
known_kinds = ('file', 'directory', 'symlink')
235
def _put_in_tar(self, item, tree):
236
"""populate item for stashing in a tar, and return the content stream.
238
If no content is available, return None.
240
raise BzrError("don't know how to export {%s} of kind %r" %
241
(self.file_id, self.kind))
243
@deprecated_method(deprecated_in((1, 6, 0)))
244
def put_on_disk(self, dest, dp, tree):
245
"""Create a representation of self on disk in the prefix dest.
247
This is a template method - implement _put_on_disk in subclasses.
249
fullpath = osutils.pathjoin(dest, dp)
250
self._put_on_disk(fullpath, tree)
251
# mutter(" export {%s} kind %s to %s", self.file_id,
252
# self.kind, fullpath)
254
def _put_on_disk(self, fullpath, tree):
255
"""Put this entry onto disk at fullpath, from tree tree."""
256
raise BzrError("don't know how to export {%s} of kind %r" % (self.file_id, self.kind))
258
def sorted_children(self):
259
return sorted(self.children.items())
262
def versionable_kind(kind):
263
return (kind in ('file', 'directory', 'symlink', 'tree-reference'))
265
def check(self, checker, rev_id, inv, tree):
266
"""Check this inventory entry is intact.
268
This is a template method, override _check for kind specific
271
:param checker: Check object providing context for the checks;
272
can be used to find out what parts of the repository have already
274
:param rev_id: Revision id from which this InventoryEntry was loaded.
275
Not necessarily the last-changed revision for this file.
276
:param inv: Inventory from which the entry was loaded.
277
:param tree: RevisionTree for this entry.
279
if self.parent_id is not None:
280
if not inv.has_id(self.parent_id):
281
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
282
% (self.parent_id, rev_id))
283
self._check(checker, rev_id, tree)
285
def _check(self, checker, rev_id, tree):
286
"""Check this inventory entry for kind specific errors."""
287
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
291
"""Clone this inventory entry."""
292
raise NotImplementedError
295
def describe_change(old_entry, new_entry):
296
"""Describe the change between old_entry and this.
298
This smells of being an InterInventoryEntry situation, but as its
299
the first one, we're making it a static method for now.
301
An entry with a different parent, or different name is considered
302
to be renamed. Reparenting is an internal detail.
303
Note that renaming the parent does not trigger a rename for the
306
# TODO: Perhaps return an object rather than just a string
307
if old_entry is new_entry:
308
# also the case of both being None
310
elif old_entry is None:
312
elif new_entry is None:
314
if old_entry.kind != new_entry.kind:
316
text_modified, meta_modified = new_entry.detect_changes(old_entry)
317
if text_modified or meta_modified:
321
# TODO 20060511 (mbp, rbc) factor out 'detect_rename' here.
322
if old_entry.parent_id != new_entry.parent_id:
324
elif old_entry.name != new_entry.name:
328
if renamed and not modified:
329
return InventoryEntry.RENAMED
330
if modified and not renamed:
332
if modified and renamed:
333
return InventoryEntry.MODIFIED_AND_RENAMED
337
return ("%s(%r, %r, parent_id=%r, revision=%r)"
338
% (self.__class__.__name__,
344
def __eq__(self, other):
346
# For the case when objects are cached
348
if not isinstance(other, InventoryEntry):
349
return NotImplemented
351
return ((self.file_id == other.file_id)
352
and (self.name == other.name)
353
and (other.symlink_target == self.symlink_target)
354
and (self.text_sha1 == other.text_sha1)
355
and (self.text_size == other.text_size)
356
and (self.text_id == other.text_id)
357
and (self.parent_id == other.parent_id)
358
and (self.kind == other.kind)
359
and (self.revision == other.revision)
360
and (self.executable == other.executable)
361
and (self.reference_revision == other.reference_revision)
364
def __ne__(self, other):
365
return not (self == other)
368
raise ValueError('not hashable')
370
def _unchanged(self, previous_ie):
371
"""Has this entry changed relative to previous_ie.
373
This method should be overridden in child classes.
376
# different inv parent
377
if previous_ie.parent_id != self.parent_id:
380
elif previous_ie.name != self.name:
382
elif previous_ie.kind != self.kind:
386
def _read_tree_state(self, path, work_tree):
387
"""Populate fields in the inventory entry from the given tree.
389
Note that this should be modified to be a noop on virtual trees
390
as all entries created there are prepopulated.
392
# TODO: Rather than running this manually, we should check the
393
# working sha1 and other expensive properties when they're
394
# first requested, or preload them if they're already known
395
pass # nothing to do by default
397
def _forget_tree_state(self):
401
class RootEntry(InventoryEntry):
403
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
404
'text_id', 'parent_id', 'children', 'executable',
405
'revision', 'symlink_target', 'reference_revision']
407
def _check(self, checker, rev_id, tree):
408
"""See InventoryEntry._check"""
410
def __init__(self, file_id):
411
self.file_id = file_id
413
self.kind = 'directory'
414
self.parent_id = None
417
symbol_versioning.warn('RootEntry is deprecated as of bzr 0.10.'
418
' Please use InventoryDirectory instead.',
419
DeprecationWarning, stacklevel=2)
421
def __eq__(self, other):
422
if not isinstance(other, RootEntry):
423
return NotImplemented
425
return (self.file_id == other.file_id) \
426
and (self.children == other.children)
429
class InventoryDirectory(InventoryEntry):
430
"""A directory in an inventory."""
432
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
433
'text_id', 'parent_id', 'children', 'executable',
434
'revision', 'symlink_target', 'reference_revision']
436
def _check(self, checker, rev_id, tree):
437
"""See InventoryEntry._check"""
438
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
439
raise BzrCheckError('directory {%s} has text in revision {%s}'
440
% (self.file_id, rev_id))
443
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
444
other.revision = self.revision
445
# note that children are *not* copied; they're pulled across when
449
def __init__(self, file_id, name, parent_id):
450
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
452
self.kind = 'directory'
454
def kind_character(self):
455
"""See InventoryEntry.kind_character."""
458
def _put_in_tar(self, item, tree):
459
"""See InventoryEntry._put_in_tar."""
460
item.type = tarfile.DIRTYPE
467
def _put_on_disk(self, fullpath, tree):
468
"""See InventoryEntry._put_on_disk."""
472
class InventoryFile(InventoryEntry):
473
"""A file in an inventory."""
475
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
476
'text_id', 'parent_id', 'children', 'executable',
477
'revision', 'symlink_target', 'reference_revision']
479
def _check(self, checker, tree_revision_id, tree):
480
"""See InventoryEntry._check"""
481
key = (self.file_id, self.revision)
482
if key in checker.checked_texts:
483
prev_sha = checker.checked_texts[key]
484
if prev_sha != self.text_sha1:
486
'mismatched sha1 on {%s} in {%s} (%s != %s) %r' %
487
(self.file_id, tree_revision_id, prev_sha, self.text_sha1,
490
checker.repeated_text_cnt += 1
493
checker.checked_text_cnt += 1
494
# We can't check the length, because Weave doesn't store that
495
# information, and the whole point of looking at the weave's
496
# sha1sum is that we don't have to extract the text.
497
if (self.text_sha1 != tree._repository.texts.get_sha1s([key])[key]):
498
raise BzrCheckError('text {%s} version {%s} wrong sha1' % key)
499
checker.checked_texts[key] = self.text_sha1
502
other = InventoryFile(self.file_id, self.name, self.parent_id)
503
other.executable = self.executable
504
other.text_id = self.text_id
125
self.text_sha1 = None
126
self.text_size = None
130
other = InventoryEntry(self.file_id, self.name, self.kind,
131
self.text_id, self.parent_id)
505
132
other.text_sha1 = self.text_sha1
506
133
other.text_size = self.text_size
507
other.revision = self.revision
510
def detect_changes(self, old_entry):
511
"""See InventoryEntry.detect_changes."""
512
text_modified = (self.text_sha1 != old_entry.text_sha1)
513
meta_modified = (self.executable != old_entry.executable)
514
return text_modified, meta_modified
516
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
517
output_to, reverse=False):
518
"""See InventoryEntry._diff."""
519
from bzrlib.diff import DiffText
520
from_file_id = self.file_id
522
to_file_id = to_entry.file_id
526
to_file_id, from_file_id = from_file_id, to_file_id
527
tree, to_tree = to_tree, tree
528
from_label, to_label = to_label, from_label
529
differ = DiffText(tree, to_tree, output_to, 'utf-8', '', '',
531
return differ.diff_text(from_file_id, to_file_id, from_label, to_label)
534
"""See InventoryEntry.has_text."""
537
def __init__(self, file_id, name, parent_id):
538
super(InventoryFile, self).__init__(file_id, name, parent_id)
541
def kind_character(self):
542
"""See InventoryEntry.kind_character."""
545
def _put_in_tar(self, item, tree):
546
"""See InventoryEntry._put_in_tar."""
547
item.type = tarfile.REGTYPE
548
fileobj = tree.get_file(self.file_id)
549
item.size = self.text_size
550
if tree.is_executable(self.file_id):
556
def _put_on_disk(self, fullpath, tree):
557
"""See InventoryEntry._put_on_disk."""
558
osutils.pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
559
if tree.is_executable(self.file_id):
560
os.chmod(fullpath, 0755)
562
def _read_tree_state(self, path, work_tree):
563
"""See InventoryEntry._read_tree_state."""
564
self.text_sha1 = work_tree.get_file_sha1(self.file_id, path=path)
565
# FIXME: 20050930 probe for the text size when getting sha1
566
# in _read_tree_state
567
self.executable = work_tree.is_executable(self.file_id, path=path)
569
137
def __repr__(self):
570
return ("%s(%r, %r, parent_id=%r, sha1=%r, len=%s, revision=%s)"
138
return ("%s(%r, %r, kind=%r, parent_id=%r)"
571
139
% (self.__class__.__name__,
579
def _forget_tree_state(self):
580
self.text_sha1 = None
582
def _unchanged(self, previous_ie):
583
"""See InventoryEntry._unchanged."""
584
compatible = super(InventoryFile, self)._unchanged(previous_ie)
585
if self.text_sha1 != previous_ie.text_sha1:
588
# FIXME: 20050930 probe for the text size when getting sha1
589
# in _read_tree_state
590
self.text_size = previous_ie.text_size
591
if self.executable != previous_ie.executable:
596
class InventoryLink(InventoryEntry):
597
"""A file in an inventory."""
599
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
600
'text_id', 'parent_id', 'children', 'executable',
601
'revision', 'symlink_target', 'reference_revision']
603
def _check(self, checker, rev_id, tree):
604
"""See InventoryEntry._check"""
605
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
606
raise BzrCheckError('symlink {%s} has text in revision {%s}'
607
% (self.file_id, rev_id))
608
if self.symlink_target is None:
609
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
610
% (self.file_id, rev_id))
613
other = InventoryLink(self.file_id, self.name, self.parent_id)
614
other.symlink_target = self.symlink_target
615
other.revision = self.revision
618
def detect_changes(self, old_entry):
619
"""See InventoryEntry.detect_changes."""
620
# FIXME: which _modified field should we use ? RBC 20051003
621
text_modified = (self.symlink_target != old_entry.symlink_target)
623
mutter(" symlink target changed")
624
meta_modified = False
625
return text_modified, meta_modified
627
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
628
output_to, reverse=False):
629
"""See InventoryEntry._diff."""
630
from bzrlib.diff import DiffSymlink
631
old_target = self.symlink_target
632
if to_entry is not None:
633
new_target = to_entry.symlink_target
642
new_target, old_target = old_target, new_target
643
differ = DiffSymlink(old_tree, new_tree, output_to)
644
return differ.diff_symlink(old_target, new_target)
646
def __init__(self, file_id, name, parent_id):
647
super(InventoryLink, self).__init__(file_id, name, parent_id)
648
self.kind = 'symlink'
650
def kind_character(self):
651
"""See InventoryEntry.kind_character."""
654
def _put_in_tar(self, item, tree):
655
"""See InventoryEntry._put_in_tar."""
656
item.type = tarfile.SYMTYPE
660
item.linkname = self.symlink_target
663
def _put_on_disk(self, fullpath, tree):
664
"""See InventoryEntry._put_on_disk."""
666
os.symlink(self.symlink_target, fullpath)
668
raise BzrError("Failed to create symlink %r -> %r, error: %s" % (fullpath, self.symlink_target, e))
670
def _read_tree_state(self, path, work_tree):
671
"""See InventoryEntry._read_tree_state."""
672
self.symlink_target = work_tree.get_symlink_target(self.file_id)
674
def _forget_tree_state(self):
675
self.symlink_target = None
677
def _unchanged(self, previous_ie):
678
"""See InventoryEntry._unchanged."""
679
compatible = super(InventoryLink, self)._unchanged(previous_ie)
680
if self.symlink_target != previous_ie.symlink_target:
685
class TreeReference(InventoryEntry):
687
kind = 'tree-reference'
689
def __init__(self, file_id, name, parent_id, revision=None,
690
reference_revision=None):
691
InventoryEntry.__init__(self, file_id, name, parent_id)
692
self.revision = revision
693
self.reference_revision = reference_revision
696
return TreeReference(self.file_id, self.name, self.parent_id,
697
self.revision, self.reference_revision)
699
def _read_tree_state(self, path, work_tree):
700
"""Populate fields in the inventory entry from the given tree.
702
self.reference_revision = work_tree.get_reference_revision(
705
def _forget_tree_state(self):
706
self.reference_revision = None
708
def _unchanged(self, previous_ie):
709
"""See InventoryEntry._unchanged."""
710
compatible = super(TreeReference, self)._unchanged(previous_ie)
711
if self.reference_revision != previous_ie.reference_revision:
716
class CommonInventory(object):
717
"""Basic inventory logic, defined in terms of primitives like has_id."""
719
def __contains__(self, file_id):
720
"""True if this entry contains a file with given id.
722
>>> inv = Inventory()
723
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
724
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
730
Note that this method along with __iter__ are not encouraged for use as
731
they are less clear than specific query methods - they may be rmeoved
734
return self.has_id(file_id)
736
def id2path(self, file_id):
737
"""Return as a string the path to file_id.
740
>>> e = i.add(InventoryDirectory('src-id', 'src', ROOT_ID))
741
>>> e = i.add(InventoryFile('foo-id', 'foo.c', parent_id='src-id'))
742
>>> print i.id2path('foo-id')
745
# get all names, skipping root
746
return '/'.join(reversed(
747
[parent.name for parent in
748
self._iter_file_id_parents(file_id)][:-1]))
750
def iter_entries(self, from_dir=None):
751
"""Return (path, entry) pairs, in order by name."""
753
if self.root is None:
757
elif isinstance(from_dir, basestring):
758
from_dir = self[from_dir]
760
# unrolling the recursive called changed the time from
761
# 440ms/663ms (inline/total) to 116ms/116ms
762
children = from_dir.children.items()
764
children = collections.deque(children)
765
stack = [(u'', children)]
767
from_dir_relpath, children = stack[-1]
770
name, ie = children.popleft()
772
# we know that from_dir_relpath never ends in a slash
773
# and 'f' doesn't begin with one, we can do a string op, rather
774
# than the checks of pathjoin(), though this means that all paths
776
path = from_dir_relpath + '/' + name
780
if ie.kind != 'directory':
783
# But do this child first
784
new_children = ie.children.items()
786
new_children = collections.deque(new_children)
787
stack.append((path, new_children))
788
# Break out of inner loop, so that we start outer loop with child
791
# if we finished all children, pop it off the stack
794
def iter_entries_by_dir(self, from_dir=None, specific_file_ids=None,
795
yield_parents=False):
796
"""Iterate over the entries in a directory first order.
798
This returns all entries for a directory before returning
799
the entries for children of a directory. This is not
800
lexicographically sorted order, and is a hybrid between
801
depth-first and breadth-first.
803
:param yield_parents: If True, yield the parents from the root leading
804
down to specific_file_ids that have been requested. This has no
805
impact if specific_file_ids is None.
806
:return: This yields (path, entry) pairs
808
if specific_file_ids and not isinstance(specific_file_ids, set):
809
specific_file_ids = set(specific_file_ids)
810
# TODO? Perhaps this should return the from_dir so that the root is
811
# yielded? or maybe an option?
813
if self.root is None:
815
# Optimize a common case
816
if (not yield_parents and specific_file_ids is not None and
817
len(specific_file_ids) == 1):
818
file_id = list(specific_file_ids)[0]
820
yield self.id2path(file_id), self[file_id]
823
if (specific_file_ids is None or yield_parents or
824
self.root.file_id in specific_file_ids):
826
elif isinstance(from_dir, basestring):
827
from_dir = self[from_dir]
829
if specific_file_ids is not None:
830
# TODO: jam 20070302 This could really be done as a loop rather
831
# than a bunch of recursive calls.
834
def add_ancestors(file_id):
835
if file_id not in byid:
837
parent_id = byid[file_id].parent_id
838
if parent_id is None:
840
if parent_id not in parents:
841
parents.add(parent_id)
842
add_ancestors(parent_id)
843
for file_id in specific_file_ids:
844
add_ancestors(file_id)
848
stack = [(u'', from_dir)]
850
cur_relpath, cur_dir = stack.pop()
853
for child_name, child_ie in sorted(cur_dir.children.iteritems()):
855
child_relpath = cur_relpath + child_name
857
if (specific_file_ids is None or
858
child_ie.file_id in specific_file_ids or
859
(yield_parents and child_ie.file_id in parents)):
860
yield child_relpath, child_ie
862
if child_ie.kind == 'directory':
863
if parents is None or child_ie.file_id in parents:
864
child_dirs.append((child_relpath+'/', child_ie))
865
stack.extend(reversed(child_dirs))
867
def _make_delta(self, old):
868
"""Make an inventory delta from two inventories."""
871
adds = new_ids - old_ids
872
deletes = old_ids - new_ids
873
common = old_ids.intersection(new_ids)
875
for file_id in deletes:
876
delta.append((old.id2path(file_id), None, file_id, None))
878
delta.append((None, self.id2path(file_id), file_id, self[file_id]))
879
for file_id in common:
880
if old[file_id] != self[file_id]:
881
delta.append((old.id2path(file_id), self.id2path(file_id),
882
file_id, self[file_id]))
885
def _get_mutable_inventory(self):
886
"""Returns a mutable copy of the object.
888
Some inventories are immutable, yet working trees, for example, needs
889
to mutate exisiting inventories instead of creating a new one.
891
raise NotImplementedError(self._get_mutable_inventory)
893
def make_entry(self, kind, name, parent_id, file_id=None):
894
"""Simple thunk to bzrlib.inventory.make_entry."""
895
return make_entry(kind, name, parent_id, file_id)
898
"""Return list of (path, ie) for all entries except the root.
900
This may be faster than iter_entries.
903
def descend(dir_ie, dir_path):
904
kids = dir_ie.children.items()
906
for name, ie in kids:
907
child_path = osutils.pathjoin(dir_path, name)
908
accum.append((child_path, ie))
909
if ie.kind == 'directory':
910
descend(ie, child_path)
912
descend(self.root, u'')
915
def directories(self):
916
"""Return (path, entry) pairs for all directories, including the root.
919
def descend(parent_ie, parent_path):
920
accum.append((parent_path, parent_ie))
922
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
925
for name, child_ie in kids:
926
child_path = osutils.pathjoin(parent_path, name)
927
descend(child_ie, child_path)
928
descend(self.root, u'')
931
def path2id(self, name):
932
"""Walk down through directories to return entry of last component.
934
names may be either a list of path components, or a single
935
string, in which case it is automatically split.
937
This returns the entry of the last component in the path,
938
which may be either a file or a directory.
940
Returns None IFF the path is not found.
942
if isinstance(name, basestring):
943
name = osutils.splitpath(name)
945
# mutter("lookup path %r" % name)
949
except errors.NoSuchId:
950
# root doesn't exist yet so nothing else can
956
children = getattr(parent, 'children', None)
965
return parent.file_id
967
def filter(self, specific_fileids):
968
"""Get an inventory view filtered against a set of file-ids.
970
Children of directories and parents are included.
972
The result may or may not reference the underlying inventory
973
so it should be treated as immutable.
975
interesting_parents = set()
976
for fileid in specific_fileids:
978
interesting_parents.update(self.get_idpath(fileid))
979
except errors.NoSuchId:
980
# This fileid is not in the inventory - that's ok
982
entries = self.iter_entries()
983
if self.root is None:
984
return Inventory(root_id=None)
985
other = Inventory(entries.next()[1].file_id)
986
other.root.revision = self.root.revision
987
other.revision_id = self.revision_id
988
directories_to_expand = set()
989
for path, entry in entries:
990
file_id = entry.file_id
991
if (file_id in specific_fileids
992
or entry.parent_id in directories_to_expand):
993
if entry.kind == 'directory':
994
directories_to_expand.add(file_id)
995
elif file_id not in interesting_parents:
997
other.add(entry.copy())
1000
def get_idpath(self, file_id):
1001
"""Return a list of file_ids for the path to an entry.
1003
The list contains one element for each directory followed by
1004
the id of the file itself. So the length of the returned list
1005
is equal to the depth of the file in the tree, counting the
1006
root directory as depth 1.
1009
for parent in self._iter_file_id_parents(file_id):
1010
p.insert(0, parent.file_id)
1014
class Inventory(CommonInventory):
146
def to_element(self):
147
"""Convert to XML element"""
150
e.set('name', self.name)
151
e.set('file_id', self.file_id)
152
e.set('kind', self.kind)
154
if self.text_size is not None:
155
e.set('text_size', '%d' % self.text_size)
157
for f in ['text_id', 'text_sha1', 'parent_id']:
167
def from_element(cls, elt):
168
assert elt.tag == 'entry'
169
self = cls(elt.get('file_id'), elt.get('name'), elt.get('kind'))
170
self.text_id = elt.get('text_id')
171
self.text_sha1 = elt.get('text_sha1')
172
self.parent_id = elt.get('parent_id')
174
## mutter("read inventoryentry: %r" % (elt.attrib))
176
v = elt.get('text_size')
177
self.text_size = v and int(v)
182
from_element = classmethod(from_element)
184
def __cmp__(self, other):
187
if not isinstance(other, InventoryEntry):
188
return NotImplemented
190
return cmp(self.file_id, other.file_id) \
191
or cmp(self.name, other.name) \
192
or cmp(self.text_sha1, other.text_sha1) \
193
or cmp(self.text_size, other.text_size) \
194
or cmp(self.text_id, other.text_id) \
195
or cmp(self.parent_id, other.parent_id) \
196
or cmp(self.kind, other.kind)
200
class Inventory(XMLMixin):
1015
201
"""Inventory of versioned files in a tree.
1017
This describes which file_id is present at each point in the tree,
1018
and possibly the SHA-1 or other information about the file.
1019
Entries can be looked up either by path or by file_id.
203
An Inventory acts like a set of InventoryEntry items. You can
204
also look files up by their file_id or name.
206
May be read from and written to a metadata file in a tree. To
207
manipulate the inventory (for example to add a file), it is read
208
in, modified, and then written back out.
1021
210
The inventory represents a typical unix file tree, with
1022
211
directories containing files and subdirectories. We never store
1044
238
May also look up by name:
1046
240
>>> [x[0] for x in inv.iter_entries()]
1048
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
1049
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
1050
Traceback (most recent call last):
1051
BzrError: parent_id {TREE_ROOT} not in inventory
1052
>>> inv.add(InventoryFile('123-123', 'hello.c', 'TREE_ROOT-12345678-12345678'))
1053
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT-12345678-12345678', sha1=None, len=None, revision=None)
243
>>> inv.write_xml(sys.stdout)
245
<entry file_id="123-123" kind="file" name="hello.c" />
1055
def __init__(self, root_id=ROOT_ID, revision_id=None):
250
## TODO: Clear up handling of files in subdirectories; we probably
251
## do want to be able to just look them up by name but this
252
## probably means gradually walking down the path, looking up as we go.
254
## TODO: Make sure only canonical filenames are stored.
256
## TODO: Do something sensible about the possible collisions on
257
## case-losing filesystems. Perhaps we should just always forbid
260
## _tree should probably just be stored as
261
## InventoryEntry._children on each directory.
1056
264
"""Create or read an inventory.
1058
266
If a working directory is specified, the inventory is read
1059
267
from there. If the file is specified, read from that. If not,
1060
268
the inventory is created empty.
1062
The inventory is created with a default root directory, with
1065
if root_id is not None:
1066
self._set_root(InventoryDirectory(root_id, u'', None))
1070
self.revision_id = revision_id
1073
# More than one page of ouput is not useful anymore to debug
1076
contents = repr(self._byid)
1077
if len(contents) > max_len:
1078
contents = contents[:(max_len-len(closing))] + closing
1079
return "<Inventory object at %x, contents=%r>" % (id(self), contents)
1081
def apply_delta(self, delta):
1082
"""Apply a delta to this inventory.
1084
:param delta: A list of changes to apply. After all the changes are
1085
applied the final inventory must be internally consistent, but it
1086
is ok to supply changes which, if only half-applied would have an
1087
invalid result - such as supplying two changes which rename two
1088
files, 'A' and 'B' with each other : [('A', 'B', 'A-id', a_entry),
1089
('B', 'A', 'B-id', b_entry)].
1091
Each change is a tuple, of the form (old_path, new_path, file_id,
1094
When new_path is None, the change indicates the removal of an entry
1095
from the inventory and new_entry will be ignored (using None is
1096
appropriate). If new_path is not None, then new_entry must be an
1097
InventoryEntry instance, which will be incorporated into the
1098
inventory (and replace any existing entry with the same file id).
1100
When old_path is None, the change indicates the addition of
1101
a new entry to the inventory.
1103
When neither new_path nor old_path are None, the change is a
1104
modification to an entry, such as a rename, reparent, kind change
1107
The children attribute of new_entry is ignored. This is because
1108
this method preserves children automatically across alterations to
1109
the parent of the children, and cases where the parent id of a
1110
child is changing require the child to be passed in as a separate
1111
change regardless. E.g. in the recursive deletion of a directory -
1112
the directory's children must be included in the delta, or the
1113
final inventory will be invalid.
1115
Note that a file_id must only appear once within a given delta.
1116
An AssertionError is raised otherwise.
1118
# Check that the delta is legal. It would be nice if this could be
1119
# done within the loops below but it's safer to validate the delta
1120
# before starting to mutate the inventory.
1121
unique_file_ids = set([f for _, _, f, _ in delta])
1122
if len(unique_file_ids) != len(delta):
1123
raise AssertionError("a file-id appears multiple times in %r"
1128
# Remove all affected items which were in the original inventory,
1129
# starting with the longest paths, thus ensuring parents are examined
1130
# after their children, which means that everything we examine has no
1131
# modified children remaining by the time we examine it.
1132
for old_path, file_id in sorted(((op, f) for op, np, f, e in delta
1133
if op is not None), reverse=True):
1134
if file_id not in self:
1137
# Preserve unaltered children of file_id for later reinsertion.
1138
file_id_children = getattr(self[file_id], 'children', {})
1139
if len(file_id_children):
1140
children[file_id] = file_id_children
1141
# Remove file_id and the unaltered children. If file_id is not
1142
# being deleted it will be reinserted back later.
1143
self.remove_recursive_id(file_id)
1144
# Insert all affected which should be in the new inventory, reattaching
1145
# their children if they had any. This is done from shortest path to
1146
# longest, ensuring that items which were modified and whose parents in
1147
# the resulting inventory were also modified, are inserted after their
1149
for new_path, new_entry in sorted((np, e) for op, np, f, e in
1150
delta if np is not None):
1151
if new_entry.kind == 'directory':
1152
# Pop the child which to allow detection of children whose
1153
# parents were deleted and which were not reattached to a new
1155
replacement = InventoryDirectory(new_entry.file_id,
1156
new_entry.name, new_entry.parent_id)
1157
replacement.revision = new_entry.revision
1158
replacement.children = children.pop(replacement.file_id, {})
1159
new_entry = replacement
1162
# Get the parent id that was deleted
1163
parent_id, children = children.popitem()
1164
raise errors.InconsistentDelta("<deleted>", parent_id,
1165
"The file id was deleted but its children were not deleted.")
1167
def _set_root(self, ie):
1169
self._byid = {self.root.file_id: self.root}
1172
# TODO: jam 20051218 Should copy also copy the revision_id?
1173
entries = self.iter_entries()
1174
if self.root is None:
1175
return Inventory(root_id=None)
1176
other = Inventory(entries.next()[1].file_id)
1177
other.root.revision = self.root.revision
1178
# copy recursively so we know directories will be added before
1179
# their children. There are more efficient ways than this...
1180
for path, entry in entries:
1181
other.add(entry.copy())
1184
def _get_mutable_inventory(self):
1185
"""See CommonInventory._get_mutable_inventory."""
1186
return deepcopy(self)
272
# _tree is indexed by parent_id; at each level a map from name
273
# to ie. The None entry is the root.
274
self._tree = {None: {}}
1188
277
def __iter__(self):
1189
"""Iterate over all file-ids."""
1190
278
return iter(self._byid)
1192
def iter_just_entries(self):
1193
"""Iterate over all entries.
1195
Unlike iter_entries(), just the entries are returned (not (path, ie))
1196
and the order of entries is undefined.
1198
XXX: We may not want to merge this into bzr.dev.
1200
if self.root is None:
1202
for _, ie in self._byid.iteritems():
1205
281
def __len__(self):
1206
282
"""Returns number of entries."""
1207
283
return len(self._byid)
286
def iter_entries(self, parent_id=None):
287
"""Return (path, entry) pairs, in order by name."""
288
kids = self._tree[parent_id].items()
290
for name, ie in kids:
292
if ie.kind == 'directory':
293
for cn, cie in self.iter_entries(parent_id=ie.file_id):
294
yield joinpath([name, cn]), cie
297
def directories(self, include_root=True):
298
"""Return (path, entry) pairs for all directories.
302
for path, entry in self.iter_entries():
303
if entry.kind == 'directory':
308
def children(self, parent_id):
309
"""Return entries that are direct children of parent_id."""
310
return self._tree[parent_id]
314
# TODO: return all paths and entries
317
def __contains__(self, file_id):
318
"""True if this entry contains a file with given id.
320
>>> inv = Inventory()
321
>>> inv.add(InventoryEntry('123', 'foo.c'))
327
return file_id in self._byid
1209
330
def __getitem__(self, file_id):
1210
331
"""Return the entry for given file_id.
1212
333
>>> inv = Inventory()
1213
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
1214
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
334
>>> inv.add(InventoryEntry('123123', 'hello.c'))
1215
335
>>> inv['123123'].name
1219
return self._byid[file_id]
1221
# really we're passing an inventory, not a tree...
1222
raise errors.NoSuchId(self, file_id)
1224
def get_file_kind(self, file_id):
1225
return self._byid[file_id].kind
1227
def get_child(self, parent_id, filename):
1228
return self[parent_id].children.get(filename)
1230
def _add_child(self, entry):
1231
"""Add an entry to the inventory, without adding it to its parent"""
1232
if entry.file_id in self._byid:
1233
raise BzrError("inventory already contains entry with id {%s}" %
1235
self._byid[entry.file_id] = entry
1236
for child in getattr(entry, 'children', {}).itervalues():
1237
self._add_child(child)
338
return self._byid[file_id]
1240
341
def add(self, entry):
1241
342
"""Add entry to inventory.
1243
344
To add a file to a branch ready to be committed, use Branch.add,
1246
Returns the new entry object.
1248
if entry.file_id in self._byid:
1249
raise errors.DuplicateFileId(entry.file_id,
1250
self._byid[entry.file_id])
1252
if entry.parent_id is None:
1256
parent = self._byid[entry.parent_id]
1258
raise BzrError("parent_id {%s} not in inventory" %
1261
if entry.name in parent.children:
1262
raise BzrError("%s is already versioned" %
1263
osutils.pathjoin(self.id2path(parent.file_id),
1264
entry.name).encode('utf-8'))
1265
parent.children[entry.name] = entry
1266
return self._add_child(entry)
1268
def add_path(self, relpath, kind, file_id=None, parent_id=None):
346
if entry.file_id in self:
347
bailout("inventory already contains entry with id {%s}" % entry.file_id)
349
if entry.parent_id != None:
350
if entry.parent_id not in self:
351
bailout("parent_id %s of new entry not found in inventory"
354
if self._tree[entry.parent_id].has_key(entry.name):
355
bailout("%s is already versioned"
356
% appendpath(self.id2path(entry.parent_id), entry.name))
358
self._byid[entry.file_id] = entry
359
self._tree[entry.parent_id][entry.name] = entry
361
if entry.kind == 'directory':
362
self._tree[entry.file_id] = {}
365
def add_path(self, relpath, kind, file_id=None):
1269
366
"""Add entry from a path.
1271
The immediate parent must already be versioned.
1273
Returns the new entry object."""
1275
parts = osutils.splitpath(relpath)
368
The immediate parent must already be versioned"""
369
parts = bzrlib.osutils.splitpath(relpath)
1277
370
if len(parts) == 0:
1279
file_id = generate_ids.gen_root_id()
1280
self.root = InventoryDirectory(file_id, '', None)
1281
self._byid = {self.root.file_id: self.root}
1284
parent_path = parts[:-1]
1285
parent_id = self.path2id(parent_path)
1286
if parent_id is None:
1287
raise errors.NotVersionedError(path=parent_path)
1288
ie = make_entry(kind, parts[-1], parent_id, file_id)
371
bailout("cannot re-add root of inventory")
374
file_id = bzrlib.branch.gen_file_id(relpath)
376
parent_id = self.path2id(parts[:-1])
377
ie = InventoryEntry(file_id, parts[-1],
378
kind=kind, parent_id=parent_id)
1289
379
return self.add(ie)
1291
382
def __delitem__(self, file_id):
1292
383
"""Remove entry by id.
1294
385
>>> inv = Inventory()
1295
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1296
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
386
>>> inv.add(InventoryEntry('123', 'foo.c'))
1297
387
>>> '123' in inv
1299
389
>>> del inv['123']
1303
393
ie = self[file_id]
395
assert self._tree[ie.parent_id][ie.name] == ie
397
# TODO: Test deleting all children; maybe hoist to a separate
399
if ie.kind == 'directory':
400
for cie in self._tree[file_id].values():
401
del self[cie.file_id]
402
del self._tree[file_id]
1304
404
del self._byid[file_id]
1305
if ie.parent_id is not None:
1306
del self[ie.parent_id].children[ie.name]
1308
def __eq__(self, other):
405
del self._tree[ie.parent_id][ie.name]
409
return Set(self._byid)
412
def to_element(self):
413
"""Convert to XML Element"""
414
e = Element('inventory')
416
for path, ie in self.iter_entries():
417
e.append(ie.to_element())
421
def from_element(cls, elt):
422
"""Construct from XML Element
424
>>> inv = Inventory()
425
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c'))
426
>>> elt = inv.to_element()
427
>>> inv2 = Inventory.from_element(elt)
431
assert elt.tag == 'inventory'
434
o.add(InventoryEntry.from_element(e))
437
from_element = classmethod(from_element)
440
def __cmp__(self, other):
1309
441
"""Compare two sets by comparing their contents.
1311
443
>>> i1 = Inventory()
1312
444
>>> i2 = Inventory()
1315
>>> i1.add(InventoryFile('123', 'foo', ROOT_ID))
1316
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
447
>>> i1.add(InventoryEntry('123', 'foo'))
1319
>>> i2.add(InventoryFile('123', 'foo', ROOT_ID))
1320
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None, revision=None)
450
>>> i2.add(InventoryEntry('123', 'foo'))
1324
457
if not isinstance(other, Inventory):
1325
458
return NotImplemented
1327
return self._byid == other._byid
1329
def __ne__(self, other):
1330
return not self.__eq__(other)
1333
raise ValueError('not hashable')
1335
def _iter_file_id_parents(self, file_id):
1336
"""Yield the parents of file_id up to the root."""
1337
while file_id is not None:
1339
ie = self._byid[file_id]
1341
raise errors.NoSuchId(tree=None, file_id=file_id)
1343
file_id = ie.parent_id
1345
def has_filename(self, filename):
1346
return bool(self.path2id(filename))
1348
def has_id(self, file_id):
1349
return (file_id in self._byid)
1351
def _make_delta(self, old):
1352
"""Make an inventory delta from two inventories."""
1353
old_getter = getattr(old, '_byid', old)
1354
new_getter = self._byid
1355
old_ids = set(old_getter)
1356
new_ids = set(new_getter)
1357
adds = new_ids - old_ids
1358
deletes = old_ids - new_ids
1359
if not adds and not deletes:
1362
common = old_ids.intersection(new_ids)
1364
for file_id in deletes:
1365
delta.append((old.id2path(file_id), None, file_id, None))
1366
for file_id in adds:
1367
delta.append((None, self.id2path(file_id), file_id, self[file_id]))
1368
for file_id in common:
1369
new_ie = new_getter[file_id]
1370
old_ie = old_getter[file_id]
1371
# If xml_serializer returns the cached InventoryEntries (rather
1372
# than always doing .copy()), inlining the 'is' check saves 2.7M
1373
# calls to __eq__. Under lsprof this saves 20s => 6s.
1374
# It is a minor improvement without lsprof.
1375
if old_ie is new_ie or old_ie == new_ie:
1378
delta.append((old.id2path(file_id), self.id2path(file_id),
1382
def remove_recursive_id(self, file_id):
1383
"""Remove file_id, and children, from the inventory.
1385
:param file_id: A file_id to remove.
1387
to_find_delete = [self._byid[file_id]]
1389
while to_find_delete:
1390
ie = to_find_delete.pop()
1391
to_delete.append(ie.file_id)
1392
if ie.kind == 'directory':
1393
to_find_delete.extend(ie.children.values())
1394
for file_id in reversed(to_delete):
460
if self.id_set() ^ other.id_set():
463
for file_id in self._byid:
464
c = cmp(self[file_id], other[file_id])
470
def id2path(self, file_id):
471
"""Return as a list the path to file_id."""
473
while file_id != None:
1395
474
ie = self[file_id]
1396
del self._byid[file_id]
1397
if ie.parent_id is not None:
1398
del self[ie.parent_id].children[ie.name]
1402
def rename(self, file_id, new_parent_id, new_name):
1403
"""Move a file within the inventory.
1405
This can change either the name, or the parent, or both.
1407
This does not move the working file.
1409
new_name = ensure_normalized_name(new_name)
1410
if not is_valid_name(new_name):
1411
raise BzrError("not an acceptable filename: %r" % new_name)
1413
new_parent = self._byid[new_parent_id]
1414
if new_name in new_parent.children:
1415
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
1417
new_parent_idpath = self.get_idpath(new_parent_id)
1418
if file_id in new_parent_idpath:
1419
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
1420
% (self.id2path(file_id), self.id2path(new_parent_id)))
1422
file_ie = self._byid[file_id]
1423
old_parent = self._byid[file_ie.parent_id]
1425
# TODO: Don't leave things messed up if this fails
1427
del old_parent.children[file_ie.name]
1428
new_parent.children[new_name] = file_ie
1430
file_ie.name = new_name
1431
file_ie.parent_id = new_parent_id
1433
def is_root(self, file_id):
1434
return self.root is not None and file_id == self.root.file_id
1437
class CHKInventory(CommonInventory):
1438
"""An inventory persisted in a CHK store.
1440
By design, a CHKInventory is immutable so many of the methods
1441
supported by Inventory - add, rename, apply_delta, etc - are *not*
1442
supported. To create a new CHKInventory, use create_by_apply_delta()
1443
or from_inventory(), say.
1445
Internally, a CHKInventory has one or two CHKMaps:
1447
* id_to_entry - a map from (file_id,) => InventoryEntry as bytes
1448
* parent_id_basename_to_file_id - a map from (parent_id, basename_utf8)
1451
The second map is optional and not present in early CHkRepository's.
1453
No caching is performed: every method call or item access will perform
1454
requests to the storage layer. As such, keep references to objects you
1458
def __init__(self, search_key_name):
1459
CommonInventory.__init__(self)
1460
self._fileid_to_entry_cache = {}
1461
self._path_to_fileid_cache = {}
1462
self._search_key_name = search_key_name
1464
def _entry_to_bytes(self, entry):
1465
"""Serialise entry as a single bytestring.
1467
:param Entry: An inventory entry.
1468
:return: A bytestring for the entry.
1471
ENTRY ::= FILE | DIR | SYMLINK | TREE
1472
FILE ::= "file: " COMMON SEP SHA SEP SIZE SEP EXECUTABLE
1473
DIR ::= "dir: " COMMON
1474
SYMLINK ::= "symlink: " COMMON SEP TARGET_UTF8
1475
TREE ::= "tree: " COMMON REFERENCE_REVISION
1476
COMMON ::= FILE_ID SEP PARENT_ID SEP NAME_UTF8 SEP REVISION
1479
if entry.parent_id is not None:
1480
parent_str = entry.parent_id
1483
name_str = entry.name.encode("utf8")
1484
if entry.kind == 'file':
1485
if entry.executable:
1489
return "file: %s\n%s\n%s\n%s\n%s\n%d\n%s" % (
1490
entry.file_id, parent_str, name_str, entry.revision,
1491
entry.text_sha1, entry.text_size, exec_str)
1492
elif entry.kind == 'directory':
1493
return "dir: %s\n%s\n%s\n%s" % (
1494
entry.file_id, parent_str, name_str, entry.revision)
1495
elif entry.kind == 'symlink':
1496
return "symlink: %s\n%s\n%s\n%s\n%s" % (
1497
entry.file_id, parent_str, name_str, entry.revision,
1498
entry.symlink_target.encode("utf8"))
1499
elif entry.kind == 'tree-reference':
1500
return "tree: %s\n%s\n%s\n%s\n%s" % (
1501
entry.file_id, parent_str, name_str, entry.revision,
1502
entry.reference_revision)
1504
raise ValueError("unknown kind %r" % entry.kind)
1507
def _bytes_to_utf8name_key(bytes):
1508
"""Get the file_id, revision_id key out of bytes."""
1509
# We don't normally care about name, except for times when we want
1510
# to filter out empty names because of non rich-root...
1511
sections = bytes.split('\n')
1512
kind, file_id = sections[0].split(': ')
1513
return (sections[2], file_id, sections[3])
1515
def _bytes_to_entry(self, bytes):
1516
"""Deserialise a serialised entry."""
1517
sections = bytes.split('\n')
1518
if sections[0].startswith("file: "):
1519
result = InventoryFile(sections[0][6:],
1520
sections[2].decode('utf8'),
1522
result.text_sha1 = sections[4]
1523
result.text_size = int(sections[5])
1524
result.executable = sections[6] == "Y"
1525
elif sections[0].startswith("dir: "):
1526
result = CHKInventoryDirectory(sections[0][5:],
1527
sections[2].decode('utf8'),
1529
elif sections[0].startswith("symlink: "):
1530
result = InventoryLink(sections[0][9:],
1531
sections[2].decode('utf8'),
1533
result.symlink_target = sections[4].decode('utf8')
1534
elif sections[0].startswith("tree: "):
1535
result = TreeReference(sections[0][6:],
1536
sections[2].decode('utf8'),
1538
result.reference_revision = sections[4]
1540
raise ValueError("Not a serialised entry %r" % bytes)
1541
result.revision = sections[3]
1542
if result.parent_id == '':
1543
result.parent_id = None
1544
self._fileid_to_entry_cache[result.file_id] = result
1547
def _get_mutable_inventory(self):
1548
"""See CommonInventory._get_mutable_inventory."""
1549
entries = self.iter_entries()
1550
if self.root_id is not None:
1552
inv = Inventory(self.root_id, self.revision_id)
1553
for path, inv_entry in entries:
1557
def create_by_apply_delta(self, inventory_delta, new_revision_id,
1558
propagate_caches=False):
1559
"""Create a new CHKInventory by applying inventory_delta to this one.
1561
:param inventory_delta: The inventory delta to apply. See
1562
Inventory.apply_delta for details.
1563
:param new_revision_id: The revision id of the resulting CHKInventory.
1564
:param propagate_caches: If True, the caches for this inventory are
1565
copied to and updated for the result.
1566
:return: The new CHKInventory.
1568
result = CHKInventory(self._search_key_name)
1569
if propagate_caches:
1570
# Just propagate the path-to-fileid cache for now
1571
result._path_to_fileid_cache = dict(self._path_to_fileid_cache.iteritems())
1572
search_key_func = chk_map.search_key_registry.get(self._search_key_name)
1573
self.id_to_entry._ensure_root()
1574
maximum_size = self.id_to_entry._root_node.maximum_size
1575
result.revision_id = new_revision_id
1576
result.id_to_entry = chk_map.CHKMap(
1577
self.id_to_entry._store,
1578
self.id_to_entry.key(),
1579
search_key_func=search_key_func)
1580
result.id_to_entry._ensure_root()
1581
result.id_to_entry._root_node.set_maximum_size(maximum_size)
1582
parent_id_basename_delta = []
1583
if self.parent_id_basename_to_file_id is not None:
1584
result.parent_id_basename_to_file_id = chk_map.CHKMap(
1585
self.parent_id_basename_to_file_id._store,
1586
self.parent_id_basename_to_file_id.key(),
1587
search_key_func=search_key_func)
1588
result.parent_id_basename_to_file_id._ensure_root()
1589
self.parent_id_basename_to_file_id._ensure_root()
1590
result_p_id_root = result.parent_id_basename_to_file_id._root_node
1591
p_id_root = self.parent_id_basename_to_file_id._root_node
1592
result_p_id_root.set_maximum_size(p_id_root.maximum_size)
1593
result_p_id_root._key_width = p_id_root._key_width
1595
result.parent_id_basename_to_file_id = None
1596
result.root_id = self.root_id
1597
id_to_entry_delta = []
1598
for old_path, new_path, file_id, entry in inventory_delta:
1601
result.root_id = file_id
1602
if new_path is None:
1607
if propagate_caches:
1609
del result._path_to_fileid_cache[old_path]
1613
new_key = (file_id,)
1614
new_value = result._entry_to_bytes(entry)
1615
# Update caches. It's worth doing this whether
1616
# we're propagating the old caches or not.
1617
result._path_to_fileid_cache[new_path] = file_id
1618
if old_path is None:
1621
old_key = (file_id,)
1622
id_to_entry_delta.append((old_key, new_key, new_value))
1623
if result.parent_id_basename_to_file_id is not None:
1624
# parent_id, basename changes
1625
if old_path is None:
1628
old_entry = self[file_id]
1629
old_key = self._parent_id_basename_key(old_entry)
1630
if new_path is None:
1634
new_key = self._parent_id_basename_key(entry)
1636
if old_key != new_key:
1637
# If the two keys are the same, the value will be unchanged
1638
# as its always the file id.
1639
parent_id_basename_delta.append((old_key, new_key, new_value))
1640
result.id_to_entry.apply_delta(id_to_entry_delta)
1641
if parent_id_basename_delta:
1642
result.parent_id_basename_to_file_id.apply_delta(parent_id_basename_delta)
1646
def deserialise(klass, chk_store, bytes, expected_revision_id):
1647
"""Deserialise a CHKInventory.
1649
:param chk_store: A CHK capable VersionedFiles instance.
1650
:param bytes: The serialised bytes.
1651
:param expected_revision_id: The revision ID we think this inventory is
1653
:return: A CHKInventory
1655
lines = bytes.split('\n')
1657
raise AssertionError('bytes to deserialize must end with an eol')
1659
if lines[0] != 'chkinventory:':
1660
raise ValueError("not a serialised CHKInventory: %r" % bytes)
1662
allowed_keys = frozenset(['root_id', 'revision_id', 'search_key_name',
1663
'parent_id_basename_to_file_id',
1665
for line in lines[1:]:
1666
key, value = line.split(': ', 1)
1667
if key not in allowed_keys:
1668
raise errors.BzrError('Unknown key in inventory: %r\n%r'
1671
raise errors.BzrError('Duplicate key in inventory: %r\n%r'
1674
revision_id = info['revision_id']
1675
root_id = info['root_id']
1676
search_key_name = info.get('search_key_name', 'plain')
1677
parent_id_basename_to_file_id = info.get(
1678
'parent_id_basename_to_file_id', None)
1679
id_to_entry = info['id_to_entry']
1681
result = CHKInventory(search_key_name)
1682
result.revision_id = revision_id
1683
result.root_id = root_id
1684
search_key_func = chk_map.search_key_registry.get(
1685
result._search_key_name)
1686
if parent_id_basename_to_file_id is not None:
1687
result.parent_id_basename_to_file_id = chk_map.CHKMap(
1688
chk_store, (parent_id_basename_to_file_id,),
1689
search_key_func=search_key_func)
1691
result.parent_id_basename_to_file_id = None
1693
result.id_to_entry = chk_map.CHKMap(chk_store, (id_to_entry,),
1694
search_key_func=search_key_func)
1695
if (result.revision_id,) != expected_revision_id:
1696
raise ValueError("Mismatched revision id and expected: %r, %r" %
1697
(result.revision_id, expected_revision_id))
1701
def from_inventory(klass, chk_store, inventory, maximum_size=0, search_key_name='plain'):
1702
"""Create a CHKInventory from an existing inventory.
1704
The content of inventory is copied into the chk_store, and a
1705
CHKInventory referencing that is returned.
1707
:param chk_store: A CHK capable VersionedFiles instance.
1708
:param inventory: The inventory to copy.
1709
:param maximum_size: The CHKMap node size limit.
1710
:param search_key_name: The identifier for the search key function
1712
result = CHKInventory(search_key_name)
1713
result.revision_id = inventory.revision_id
1714
result.root_id = inventory.root.file_id
1715
search_key_func = chk_map.search_key_registry.get(search_key_name)
1716
result.id_to_entry = chk_map.CHKMap(chk_store, None, search_key_func)
1717
result.id_to_entry._root_node.set_maximum_size(maximum_size)
1719
result.parent_id_basename_to_file_id = chk_map.CHKMap(chk_store,
1720
None, search_key_func)
1721
result.parent_id_basename_to_file_id._root_node.set_maximum_size(
1723
result.parent_id_basename_to_file_id._root_node._key_width = 2
1724
parent_id_delta = []
1725
for path, entry in inventory.iter_entries():
1726
file_id_delta.append((None, (entry.file_id,),
1727
result._entry_to_bytes(entry)))
1728
parent_id_delta.append(
1729
(None, result._parent_id_basename_key(entry),
1731
result.id_to_entry.apply_delta(file_id_delta)
1732
result.parent_id_basename_to_file_id.apply_delta(parent_id_delta)
1735
def _parent_id_basename_key(self, entry):
1736
"""Create a key for a entry in a parent_id_basename_to_file_id index."""
1737
if entry.parent_id is not None:
1738
parent_id = entry.parent_id
1741
return parent_id, entry.name.encode('utf8')
1743
def __getitem__(self, file_id):
1744
"""map a single file_id -> InventoryEntry."""
1746
raise errors.NoSuchId(self, file_id)
1747
result = self._fileid_to_entry_cache.get(file_id, None)
1748
if result is not None:
1751
return self._bytes_to_entry(
1752
self.id_to_entry.iteritems([(file_id,)]).next()[1])
1753
except StopIteration:
1754
# really we're passing an inventory, not a tree...
1755
raise errors.NoSuchId(self, file_id)
1757
def has_id(self, file_id):
1758
# Perhaps have an explicit 'contains' method on CHKMap ?
1759
if self._fileid_to_entry_cache.get(file_id, None) is not None:
1761
return len(list(self.id_to_entry.iteritems([(file_id,)]))) == 1
1763
def is_root(self, file_id):
1764
return file_id == self.root_id
1766
def _iter_file_id_parents(self, file_id):
1767
"""Yield the parents of file_id up to the root."""
1768
while file_id is not None:
476
file_id = ie.parent_id
481
def path2id(self, name):
482
"""Walk down through directories to return entry of last component.
484
names may be either a list of path components, or a single
485
string, in which case it is automatically split.
487
This returns the entry of the last component in the path,
488
which may be either a file or a directory.
490
if isinstance(name, types.StringTypes):
491
name = splitpath(name)
496
cie = self._tree[parent_id][f]
498
parent_id = cie.file_id
1771
499
except KeyError:
1772
raise errors.NoSuchId(tree=self, file_id=file_id)
1774
file_id = ie.parent_id
1777
"""Iterate over all file-ids."""
1778
for key, _ in self.id_to_entry.iteritems():
1781
def iter_just_entries(self):
1782
"""Iterate over all entries.
1784
Unlike iter_entries(), just the entries are returned (not (path, ie))
1785
and the order of entries is undefined.
1787
XXX: We may not want to merge this into bzr.dev.
1789
for key, entry in self.id_to_entry.iteritems():
1791
ie = self._fileid_to_entry_cache.get(file_id, None)
1793
ie = self._bytes_to_entry(entry)
1794
self._fileid_to_entry_cache[file_id] = ie
1797
def iter_changes(self, basis):
1798
"""Generate a Tree.iter_changes change list between this and basis.
1800
:param basis: Another CHKInventory.
1801
:return: An iterator over the changes between self and basis, as per
1802
tree.iter_changes().
1804
# We want: (file_id, (path_in_source, path_in_target),
1805
# changed_content, versioned, parent, name, kind,
1807
for key, basis_value, self_value in \
1808
self.id_to_entry.iter_changes(basis.id_to_entry):
1810
if basis_value is not None:
1811
basis_entry = basis._bytes_to_entry(basis_value)
1812
path_in_source = basis.id2path(file_id)
1813
basis_parent = basis_entry.parent_id
1814
basis_name = basis_entry.name
1815
basis_executable = basis_entry.executable
1817
path_in_source = None
1820
basis_executable = None
1821
if self_value is not None:
1822
self_entry = self._bytes_to_entry(self_value)
1823
path_in_target = self.id2path(file_id)
1824
self_parent = self_entry.parent_id
1825
self_name = self_entry.name
1826
self_executable = self_entry.executable
1828
path_in_target = None
1831
self_executable = None
1832
if basis_value is None:
1834
kind = (None, self_entry.kind)
1835
versioned = (False, True)
1836
elif self_value is None:
1838
kind = (basis_entry.kind, None)
1839
versioned = (True, False)
1841
kind = (basis_entry.kind, self_entry.kind)
1842
versioned = (True, True)
1843
changed_content = False
1844
if kind[0] != kind[1]:
1845
changed_content = True
1846
elif kind[0] == 'file':
1847
if (self_entry.text_size != basis_entry.text_size or
1848
self_entry.text_sha1 != basis_entry.text_sha1):
1849
changed_content = True
1850
elif kind[0] == 'symlink':
1851
if self_entry.symlink_target != basis_entry.symlink_target:
1852
changed_content = True
1853
elif kind[0] == 'tree-reference':
1854
if (self_entry.reference_revision !=
1855
basis_entry.reference_revision):
1856
changed_content = True
1857
parent = (basis_parent, self_parent)
1858
name = (basis_name, self_name)
1859
executable = (basis_executable, self_executable)
1860
if (not changed_content
1861
and parent[0] == parent[1]
1862
and name[0] == name[1]
1863
and executable[0] == executable[1]):
1864
# Could happen when only the revision changed for a directory
1867
yield (file_id, (path_in_source, path_in_target), changed_content,
1868
versioned, parent, name, kind, executable)
1871
"""Return the number of entries in the inventory."""
1872
return len(self.id_to_entry)
1874
def _make_delta(self, old):
1875
"""Make an inventory delta from two inventories."""
1876
if type(old) != CHKInventory:
1877
return CommonInventory._make_delta(self, old)
1879
for key, old_value, self_value in \
1880
self.id_to_entry.iter_changes(old.id_to_entry):
1882
if old_value is not None:
1883
old_path = old.id2path(file_id)
1886
if self_value is not None:
1887
entry = self._bytes_to_entry(self_value)
1888
self._fileid_to_entry_cache[file_id] = entry
1889
new_path = self.id2path(file_id)
1893
delta.append((old_path, new_path, file_id, entry))
1896
def path2id(self, name):
1897
"""See CommonInventory.path2id()."""
1898
result = self._path_to_fileid_cache.get(name, None)
1900
result = CommonInventory.path2id(self, name)
1901
self._path_to_fileid_cache[name] = result
1905
"""Serialise the inventory to lines."""
1906
lines = ["chkinventory:\n"]
1907
if self._search_key_name != 'plain':
1908
# custom ordering grouping things that don't change together
1909
lines.append('search_key_name: %s\n' % (self._search_key_name,))
1910
lines.append("root_id: %s\n" % self.root_id)
1911
lines.append('parent_id_basename_to_file_id: %s\n' %
1912
self.parent_id_basename_to_file_id.key())
1913
lines.append("revision_id: %s\n" % self.revision_id)
1914
lines.append("id_to_entry: %s\n" % self.id_to_entry.key())
1916
lines.append("revision_id: %s\n" % self.revision_id)
1917
lines.append("root_id: %s\n" % self.root_id)
1918
if self.parent_id_basename_to_file_id is not None:
1919
lines.append('parent_id_basename_to_file_id: %s\n' %
1920
self.parent_id_basename_to_file_id.key())
1921
lines.append("id_to_entry: %s\n" % self.id_to_entry.key())
1926
"""Get the root entry."""
1927
return self[self.root_id]
1930
class CHKInventoryDirectory(InventoryDirectory):
1931
"""A directory in an inventory."""
1933
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
1934
'text_id', 'parent_id', '_children', 'executable',
1935
'revision', 'symlink_target', 'reference_revision',
1938
def __init__(self, file_id, name, parent_id, chk_inventory):
1939
# Don't call InventoryDirectory.__init__ - it isn't right for this
1941
InventoryEntry.__init__(self, file_id, name, parent_id)
1942
self._children = None
1943
self.kind = 'directory'
1944
self._chk_inventory = chk_inventory
1948
"""Access the list of children of this directory.
1950
With a parent_id_basename_to_file_id index, loads all the children,
1951
without loads the entire index. Without is bad. A more sophisticated
1952
proxy object might be nice, to allow partial loading of children as
1953
well when specific names are accessed. (So path traversal can be
1954
written in the obvious way but not examine siblings.).
1956
if self._children is not None:
1957
return self._children
1958
# No longer supported
1959
if self._chk_inventory.parent_id_basename_to_file_id is None:
1960
raise AssertionError("Inventories without"
1961
" parent_id_basename_to_file_id are no longer supported")
1963
# XXX: Todo - use proxy objects for the children rather than loading
1964
# all when the attribute is referenced.
1965
parent_id_index = self._chk_inventory.parent_id_basename_to_file_id
1967
for (parent_id, name_utf8), file_id in parent_id_index.iteritems(
1968
key_filter=[(self.file_id,)]):
1969
child_keys.add((file_id,))
1971
for file_id_key in child_keys:
1972
entry = self._chk_inventory._fileid_to_entry_cache.get(
1973
file_id_key[0], None)
1974
if entry is not None:
1975
result[entry.name] = entry
1976
cached.add(file_id_key)
1977
child_keys.difference_update(cached)
1978
# populate; todo: do by name
1979
id_to_entry = self._chk_inventory.id_to_entry
1980
for file_id_key, bytes in id_to_entry.iteritems(child_keys):
1981
entry = self._chk_inventory._bytes_to_entry(bytes)
1982
result[entry.name] = entry
1983
self._chk_inventory._fileid_to_entry_cache[file_id_key[0]] = entry
1984
self._children = result
1988
'directory': InventoryDirectory,
1989
'file': InventoryFile,
1990
'symlink': InventoryLink,
1991
'tree-reference': TreeReference
1994
def make_entry(kind, name, parent_id, file_id=None):
1995
"""Create an inventory entry.
1997
:param kind: the type of inventory entry to create.
1998
:param name: the basename of the entry.
1999
:param parent_id: the parent_id of the entry.
2000
:param file_id: the file_id to use. if None, one will be created.
2003
file_id = generate_ids.gen_file_id(name)
2004
name = ensure_normalized_name(name)
2006
factory = entry_factory[kind]
2008
raise BzrError("unknown kind %r" % kind)
2009
return factory(file_id, name, parent_id)
2012
def ensure_normalized_name(name):
2015
:raises InvalidNormalization: When name is not normalized, and cannot be
2016
accessed on this platform by the normalized path.
2017
:return: The NFC normalised version of name.
2019
#------- This has been copied to bzrlib.dirstate.DirState.add, please
2020
# keep them synchronised.
2021
# we dont import normalized_filename directly because we want to be
2022
# able to change the implementation at runtime for tests.
2023
norm_name, can_access = osutils.normalized_filename(name)
2024
if norm_name != name:
2028
# TODO: jam 20060701 This would probably be more useful
2029
# if the error was raised with the full path
2030
raise errors.InvalidNormalization(name)
2036
def is_valid_name(name):
2038
if _NAME_RE is None:
2039
_NAME_RE = re.compile(r'^[^/\\]+$')
2041
return bool(_NAME_RE.match(name))
506
def get_child(self, parent_id, child_name):
507
return self._tree[parent_id].get(child_name)
510
def has_filename(self, names):
511
return bool(self.path2id(names))
514
def has_id(self, file_id):
515
assert isinstance(file_id, str)
516
return self._byid.has_key(file_id)
522
if __name__ == '__main__':
523
import doctest, inventory
524
doctest.testmod(inventory)
added
removed
bzrlib/inventory.py
