2
# -*- coding: UTF-8 -*-
1
# Copyright (C) 2005, 2006, 2007 Canonical Ltd
4
3
# This program is free software; you can redistribute it and/or modify
5
4
# it under the terms of the GNU General Public License as published by
6
5
# the Free Software Foundation; either version 2 of the License, or
7
6
# (at your option) any later version.
9
8
# This program is distributed in the hope that it will be useful,
10
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
11
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12
11
# GNU General Public License for more details.
14
13
# You should have received a copy of the GNU General Public License
15
14
# along with this program; if not, write to the Free Software
16
15
# 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
34
from osutils import uuid, quotefn, splitpath, joinpath, appendpath
35
from trace import mutter
37
class InventoryEntry(XMLMixin):
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.
34
from bzrlib.lazy_import import lazy_import
35
lazy_import(globals(), """
49
from bzrlib.errors import (
53
from bzrlib.symbol_versioning import deprecated_in, deprecated_method
54
from bzrlib.trace import mutter
57
class InventoryEntry(object):
38
58
"""Description of a versioned file.
40
60
An InventoryEntry has the following fields, which are also
41
61
present in the XML inventory-entry element:
44
* *name*: (only the basename within the directory, must not
46
* *kind*: "directory" or "file"
47
* *directory_id*: (if absent/null means the branch root directory)
48
* *text_sha1*: only for files
49
* *text_size*: in bytes, only for files
50
* *text_id*: identifier for the text version, only for files
52
InventoryEntries can also exist inside a WorkingTree
53
inventory, in which case they are not yet bound to a
54
particular revision of the file. In that case the text_sha1,
55
text_size and text_id are absent.
66
(within the parent directory)
69
file_id of the parent directory, or ROOT_ID
72
the revision_id in which this variation of this file was
76
Indicates that this file should be executable on systems
80
sha-1 of the text of the file
83
size in bytes of the text of the file
85
(reading a version 4 tree created a text_id field.)
58
87
>>> i = Inventory()
60
>>> i.add(InventoryEntry('123', 'src', kind='directory'))
61
>>> i.add(InventoryEntry('2323', 'hello.c', parent_id='123'))
62
>>> for j in i.iter_entries():
90
>>> i.add(InventoryDirectory('123', 'src', ROOT_ID))
91
InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None)
92
>>> i.add(InventoryFile('2323', 'hello.c', parent_id='123'))
93
InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None)
94
>>> shouldbe = {0: '', 1: 'src', 2: 'src/hello.c'}
95
>>> for ix, j in enumerate(i.iter_entries()):
96
... print (j[0] == shouldbe[ix], j[1])
65
('src', InventoryEntry('123', 'src', kind='directory', parent_id=None))
66
('src/hello.c', InventoryEntry('2323', 'hello.c', kind='file', parent_id='123'))
67
>>> i.add(InventoryEntry('2323', 'bye.c', parent_id='123'))
68
Traceback (most recent call last):
70
BzrError: ('inventory already contains entry with id {2323}', [])
71
>>> i.add(InventoryEntry('2324', 'bye.c', parent_id='123'))
72
>>> i.add(InventoryEntry('2325', 'wibble', parent_id='123', kind='directory'))
98
(True, InventoryDirectory('TREE_ROOT', u'', parent_id=None, revision=None))
99
(True, InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None))
100
(True, InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None))
101
>>> i.add(InventoryFile('2324', 'bye.c', '123'))
102
InventoryFile('2324', 'bye.c', parent_id='123', sha1=None, len=None)
103
>>> i.add(InventoryDirectory('2325', 'wibble', '123'))
104
InventoryDirectory('2325', 'wibble', parent_id='123', revision=None)
73
105
>>> i.path2id('src/wibble')
77
>>> i.add(InventoryEntry('2326', 'wibble.c', parent_id='2325'))
109
>>> i.add(InventoryFile('2326', 'wibble.c', '2325'))
110
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
79
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
80
>>> for j in i.iter_entries():
82
... assert i.path2id(j[0])
112
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
113
>>> for path, entry in i.iter_entries():
88
121
src/wibble/wibble.c
89
122
>>> i.id2path('2326')
90
123
'src/wibble/wibble.c'
92
:todo: Maybe also keep the full path of the entry, and the children?
93
But those depend on its position within a particular inventory, and
94
it would be nice not to need to hold the backpointer here.
96
def __init__(self, file_id, name, kind='file', text_id=None,
126
# Constants returned by describe_change()
128
# TODO: These should probably move to some kind of FileChangeDescription
129
# class; that's like what's inside a TreeDelta but we want to be able to
130
# generate them just for one file at a time.
132
MODIFIED_AND_RENAMED = 'modified and renamed'
136
def detect_changes(self, old_entry):
137
"""Return a (text_modified, meta_modified) from this to old_entry.
139
_read_tree_state must have been called on self and old_entry prior to
140
calling detect_changes.
144
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
145
output_to, reverse=False):
146
"""Perform a diff between two entries of the same kind."""
148
def parent_candidates(self, previous_inventories):
149
"""Find possible per-file graph parents.
151
This is currently defined by:
152
- Select the last changed revision in the parent inventory.
153
- Do deal with a short lived bug in bzr 0.8's development two entries
154
that have the same last changed but different 'x' bit settings are
157
# revision:ie mapping for each ie found in previous_inventories.
159
# identify candidate head revision ids.
160
for inv in previous_inventories:
161
if self.file_id in inv:
162
ie = inv[self.file_id]
163
if ie.revision in candidates:
164
# same revision value in two different inventories:
165
# correct possible inconsistencies:
166
# * there was a bug in revision updates with 'x' bit
169
if candidates[ie.revision].executable != ie.executable:
170
candidates[ie.revision].executable = False
171
ie.executable = False
172
except AttributeError:
175
# add this revision as a candidate.
176
candidates[ie.revision] = ie
179
@deprecated_method(deprecated_in((1, 6, 0)))
180
def get_tar_item(self, root, dp, now, tree):
181
"""Get a tarfile item and a file stream for its content."""
182
item = tarfile.TarInfo(osutils.pathjoin(root, dp).encode('utf8'))
183
# TODO: would be cool to actually set it to the timestamp of the
184
# revision it was last changed
186
fileobj = self._put_in_tar(item, tree)
190
"""Return true if the object this entry represents has textual data.
192
Note that textual data includes binary content.
194
Also note that all entries get weave files created for them.
195
This attribute is primarily used when upgrading from old trees that
196
did not have the weave index for all inventory entries.
200
def __init__(self, file_id, name, parent_id, text_id=None):
98
201
"""Create an InventoryEntry
100
203
The filename must be a single component, relative to the
101
204
parent directory; it cannot be a whole path or relative name.
103
>>> e = InventoryEntry('123', 'hello.c')
206
>>> e = InventoryFile('123', 'hello.c', ROOT_ID)
108
>>> e = InventoryEntry('123', 'src/hello.c')
211
>>> e = InventoryFile('123', 'src/hello.c', ROOT_ID)
109
212
Traceback (most recent call last):
110
BzrError: ("InventoryEntry name is not a simple filename: 'src/hello.c'", [])
213
InvalidEntryName: Invalid entry name: src/hello.c
113
if len(splitpath(name)) != 1:
114
bailout('InventoryEntry name is not a simple filename: %r'
215
if '/' in name or '\\' in name:
216
raise errors.InvalidEntryName(name=name)
217
self.executable = False
219
self.text_sha1 = None
220
self.text_size = None
117
221
self.file_id = file_id
119
assert kind in ['file', 'directory']
121
223
self.text_id = text_id
122
224
self.parent_id = parent_id
123
self.text_sha1 = None
124
self.text_size = None
128
other = InventoryEntry(self.file_id, self.name, self.kind,
129
self.text_id, self.parent_id)
225
self.symlink_target = None
226
self.reference_revision = None
228
def kind_character(self):
229
"""Return a short kind indicator useful for appending to names."""
230
raise BzrError('unknown kind %r' % self.kind)
232
known_kinds = ('file', 'directory', 'symlink')
234
def _put_in_tar(self, item, tree):
235
"""populate item for stashing in a tar, and return the content stream.
237
If no content is available, return None.
239
raise BzrError("don't know how to export {%s} of kind %r" %
240
(self.file_id, self.kind))
242
@deprecated_method(deprecated_in((1, 6, 0)))
243
def put_on_disk(self, dest, dp, tree):
244
"""Create a representation of self on disk in the prefix dest.
246
This is a template method - implement _put_on_disk in subclasses.
248
fullpath = osutils.pathjoin(dest, dp)
249
self._put_on_disk(fullpath, tree)
250
# mutter(" export {%s} kind %s to %s", self.file_id,
251
# self.kind, fullpath)
253
def _put_on_disk(self, fullpath, tree):
254
"""Put this entry onto disk at fullpath, from tree tree."""
255
raise BzrError("don't know how to export {%s} of kind %r" % (self.file_id, self.kind))
257
def sorted_children(self):
258
return sorted(self.children.items())
261
def versionable_kind(kind):
262
return (kind in ('file', 'directory', 'symlink', 'tree-reference'))
264
def check(self, checker, rev_id, inv, tree):
265
"""Check this inventory entry is intact.
267
This is a template method, override _check for kind specific
270
:param checker: Check object providing context for the checks;
271
can be used to find out what parts of the repository have already
273
:param rev_id: Revision id from which this InventoryEntry was loaded.
274
Not necessarily the last-changed revision for this file.
275
:param inv: Inventory from which the entry was loaded.
276
:param tree: RevisionTree for this entry.
278
if self.parent_id is not None:
279
if not inv.has_id(self.parent_id):
280
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
281
% (self.parent_id, rev_id))
282
self._check(checker, rev_id, tree)
284
def _check(self, checker, rev_id, tree):
285
"""Check this inventory entry for kind specific errors."""
286
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
290
"""Clone this inventory entry."""
291
raise NotImplementedError
294
def describe_change(old_entry, new_entry):
295
"""Describe the change between old_entry and this.
297
This smells of being an InterInventoryEntry situation, but as its
298
the first one, we're making it a static method for now.
300
An entry with a different parent, or different name is considered
301
to be renamed. Reparenting is an internal detail.
302
Note that renaming the parent does not trigger a rename for the
305
# TODO: Perhaps return an object rather than just a string
306
if old_entry is new_entry:
307
# also the case of both being None
309
elif old_entry is None:
311
elif new_entry is None:
313
if old_entry.kind != new_entry.kind:
315
text_modified, meta_modified = new_entry.detect_changes(old_entry)
316
if text_modified or meta_modified:
320
# TODO 20060511 (mbp, rbc) factor out 'detect_rename' here.
321
if old_entry.parent_id != new_entry.parent_id:
323
elif old_entry.name != new_entry.name:
327
if renamed and not modified:
328
return InventoryEntry.RENAMED
329
if modified and not renamed:
331
if modified and renamed:
332
return InventoryEntry.MODIFIED_AND_RENAMED
336
return ("%s(%r, %r, parent_id=%r, revision=%r)"
337
% (self.__class__.__name__,
343
def __eq__(self, other):
344
if not isinstance(other, InventoryEntry):
345
return NotImplemented
347
return ((self.file_id == other.file_id)
348
and (self.name == other.name)
349
and (other.symlink_target == self.symlink_target)
350
and (self.text_sha1 == other.text_sha1)
351
and (self.text_size == other.text_size)
352
and (self.text_id == other.text_id)
353
and (self.parent_id == other.parent_id)
354
and (self.kind == other.kind)
355
and (self.revision == other.revision)
356
and (self.executable == other.executable)
357
and (self.reference_revision == other.reference_revision)
360
def __ne__(self, other):
361
return not (self == other)
364
raise ValueError('not hashable')
366
def _unchanged(self, previous_ie):
367
"""Has this entry changed relative to previous_ie.
369
This method should be overridden in child classes.
372
# different inv parent
373
if previous_ie.parent_id != self.parent_id:
376
elif previous_ie.name != self.name:
378
elif previous_ie.kind != self.kind:
382
def _read_tree_state(self, path, work_tree):
383
"""Populate fields in the inventory entry from the given tree.
385
Note that this should be modified to be a noop on virtual trees
386
as all entries created there are prepopulated.
388
# TODO: Rather than running this manually, we should check the
389
# working sha1 and other expensive properties when they're
390
# first requested, or preload them if they're already known
391
pass # nothing to do by default
393
def _forget_tree_state(self):
397
class RootEntry(InventoryEntry):
399
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
400
'text_id', 'parent_id', 'children', 'executable',
401
'revision', 'symlink_target', 'reference_revision']
403
def _check(self, checker, rev_id, tree):
404
"""See InventoryEntry._check"""
406
def __init__(self, file_id):
407
self.file_id = file_id
409
self.kind = 'directory'
410
self.parent_id = None
413
symbol_versioning.warn('RootEntry is deprecated as of bzr 0.10.'
414
' Please use InventoryDirectory instead.',
415
DeprecationWarning, stacklevel=2)
417
def __eq__(self, other):
418
if not isinstance(other, RootEntry):
419
return NotImplemented
421
return (self.file_id == other.file_id) \
422
and (self.children == other.children)
425
class InventoryDirectory(InventoryEntry):
426
"""A directory in an inventory."""
428
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
429
'text_id', 'parent_id', 'children', 'executable',
430
'revision', 'symlink_target', 'reference_revision']
432
def _check(self, checker, rev_id, tree):
433
"""See InventoryEntry._check"""
434
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
435
raise BzrCheckError('directory {%s} has text in revision {%s}'
436
% (self.file_id, rev_id))
439
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
440
other.revision = self.revision
441
# note that children are *not* copied; they're pulled across when
445
def __init__(self, file_id, name, parent_id):
446
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
448
self.kind = 'directory'
450
def kind_character(self):
451
"""See InventoryEntry.kind_character."""
454
def _put_in_tar(self, item, tree):
455
"""See InventoryEntry._put_in_tar."""
456
item.type = tarfile.DIRTYPE
463
def _put_on_disk(self, fullpath, tree):
464
"""See InventoryEntry._put_on_disk."""
468
class InventoryFile(InventoryEntry):
469
"""A file in an inventory."""
471
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
472
'text_id', 'parent_id', 'children', 'executable',
473
'revision', 'symlink_target', 'reference_revision']
475
def _check(self, checker, tree_revision_id, tree):
476
"""See InventoryEntry._check"""
477
key = (self.file_id, self.revision)
478
if key in checker.checked_texts:
479
prev_sha = checker.checked_texts[key]
480
if prev_sha != self.text_sha1:
482
'mismatched sha1 on {%s} in {%s} (%s != %s) %r' %
483
(self.file_id, tree_revision_id, prev_sha, self.text_sha1,
486
checker.repeated_text_cnt += 1
489
mutter('check version {%s} of {%s}', tree_revision_id, self.file_id)
490
checker.checked_text_cnt += 1
491
# We can't check the length, because Weave doesn't store that
492
# information, and the whole point of looking at the weave's
493
# sha1sum is that we don't have to extract the text.
494
if (self.text_sha1 != tree._repository.texts.get_sha1s([key])[key]):
495
raise BzrCheckError('text {%s} version {%s} wrong sha1' % key)
496
checker.checked_texts[key] = self.text_sha1
499
other = InventoryFile(self.file_id, self.name, self.parent_id)
500
other.executable = self.executable
501
other.text_id = self.text_id
130
502
other.text_sha1 = self.text_sha1
131
503
other.text_size = self.text_size
504
other.revision = self.revision
507
def detect_changes(self, old_entry):
508
"""See InventoryEntry.detect_changes."""
509
text_modified = (self.text_sha1 != old_entry.text_sha1)
510
meta_modified = (self.executable != old_entry.executable)
511
return text_modified, meta_modified
513
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
514
output_to, reverse=False):
515
"""See InventoryEntry._diff."""
516
from bzrlib.diff import DiffText
517
from_file_id = self.file_id
519
to_file_id = to_entry.file_id
523
to_file_id, from_file_id = from_file_id, to_file_id
524
tree, to_tree = to_tree, tree
525
from_label, to_label = to_label, from_label
526
differ = DiffText(tree, to_tree, output_to, 'utf-8', '', '',
528
return differ.diff_text(from_file_id, to_file_id, from_label, to_label)
531
"""See InventoryEntry.has_text."""
534
def __init__(self, file_id, name, parent_id):
535
super(InventoryFile, self).__init__(file_id, name, parent_id)
538
def kind_character(self):
539
"""See InventoryEntry.kind_character."""
542
def _put_in_tar(self, item, tree):
543
"""See InventoryEntry._put_in_tar."""
544
item.type = tarfile.REGTYPE
545
fileobj = tree.get_file(self.file_id)
546
item.size = self.text_size
547
if tree.is_executable(self.file_id):
553
def _put_on_disk(self, fullpath, tree):
554
"""See InventoryEntry._put_on_disk."""
555
osutils.pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
556
if tree.is_executable(self.file_id):
557
os.chmod(fullpath, 0755)
559
def _read_tree_state(self, path, work_tree):
560
"""See InventoryEntry._read_tree_state."""
561
self.text_sha1 = work_tree.get_file_sha1(self.file_id, path=path)
562
# FIXME: 20050930 probe for the text size when getting sha1
563
# in _read_tree_state
564
self.executable = work_tree.is_executable(self.file_id, path=path)
135
566
def __repr__(self):
136
return ("%s(%r, %r, kind=%r, parent_id=%r)"
567
return ("%s(%r, %r, parent_id=%r, sha1=%r, len=%s)"
137
568
% (self.__class__.__name__,
144
def to_element(self):
145
"""Convert to XML element"""
148
e.set('name', self.name)
149
e.set('file_id', self.file_id)
150
e.set('kind', self.kind)
152
if self.text_size is not None:
153
e.set('text_size', '%d' % self.text_size)
155
for f in ['text_id', 'text_sha1', 'parent_id']:
165
def from_element(cls, elt):
166
assert elt.tag == 'entry'
167
self = cls(elt.get('file_id'), elt.get('name'), elt.get('kind'))
168
self.text_id = elt.get('text_id')
169
self.text_sha1 = elt.get('text_sha1')
170
self.parent_id = elt.get('parent_id')
172
## mutter("read inventoryentry: %r" % (elt.attrib))
174
v = elt.get('text_size')
175
self.text_size = v and int(v)
180
from_element = classmethod(from_element)
182
def __cmp__(self, other):
185
if not isinstance(other, InventoryEntry):
186
return NotImplemented
188
return cmp(self.file_id, other.file_id) \
189
or cmp(self.name, other.name) \
190
or cmp(self.text_sha1, other.text_sha1) \
191
or cmp(self.text_size, other.text_size) \
192
or cmp(self.text_id, other.text_id) \
193
or cmp(self.parent_id, other.parent_id) \
194
or cmp(self.kind, other.kind)
198
class Inventory(XMLMixin):
575
def _forget_tree_state(self):
576
self.text_sha1 = None
578
def _unchanged(self, previous_ie):
579
"""See InventoryEntry._unchanged."""
580
compatible = super(InventoryFile, self)._unchanged(previous_ie)
581
if self.text_sha1 != previous_ie.text_sha1:
584
# FIXME: 20050930 probe for the text size when getting sha1
585
# in _read_tree_state
586
self.text_size = previous_ie.text_size
587
if self.executable != previous_ie.executable:
592
class InventoryLink(InventoryEntry):
593
"""A file in an inventory."""
595
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
596
'text_id', 'parent_id', 'children', 'executable',
597
'revision', 'symlink_target', 'reference_revision']
599
def _check(self, checker, rev_id, tree):
600
"""See InventoryEntry._check"""
601
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
602
raise BzrCheckError('symlink {%s} has text in revision {%s}'
603
% (self.file_id, rev_id))
604
if self.symlink_target is None:
605
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
606
% (self.file_id, rev_id))
609
other = InventoryLink(self.file_id, self.name, self.parent_id)
610
other.symlink_target = self.symlink_target
611
other.revision = self.revision
614
def detect_changes(self, old_entry):
615
"""See InventoryEntry.detect_changes."""
616
# FIXME: which _modified field should we use ? RBC 20051003
617
text_modified = (self.symlink_target != old_entry.symlink_target)
619
mutter(" symlink target changed")
620
meta_modified = False
621
return text_modified, meta_modified
623
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
624
output_to, reverse=False):
625
"""See InventoryEntry._diff."""
626
from bzrlib.diff import DiffSymlink
627
old_target = self.symlink_target
628
if to_entry is not None:
629
new_target = to_entry.symlink_target
638
new_target, old_target = old_target, new_target
639
differ = DiffSymlink(old_tree, new_tree, output_to)
640
return differ.diff_symlink(old_target, new_target)
642
def __init__(self, file_id, name, parent_id):
643
super(InventoryLink, self).__init__(file_id, name, parent_id)
644
self.kind = 'symlink'
646
def kind_character(self):
647
"""See InventoryEntry.kind_character."""
650
def _put_in_tar(self, item, tree):
651
"""See InventoryEntry._put_in_tar."""
652
item.type = tarfile.SYMTYPE
656
item.linkname = self.symlink_target
659
def _put_on_disk(self, fullpath, tree):
660
"""See InventoryEntry._put_on_disk."""
662
os.symlink(self.symlink_target, fullpath)
664
raise BzrError("Failed to create symlink %r -> %r, error: %s" % (fullpath, self.symlink_target, e))
666
def _read_tree_state(self, path, work_tree):
667
"""See InventoryEntry._read_tree_state."""
668
self.symlink_target = work_tree.get_symlink_target(self.file_id)
670
def _forget_tree_state(self):
671
self.symlink_target = None
673
def _unchanged(self, previous_ie):
674
"""See InventoryEntry._unchanged."""
675
compatible = super(InventoryLink, self)._unchanged(previous_ie)
676
if self.symlink_target != previous_ie.symlink_target:
681
class TreeReference(InventoryEntry):
683
kind = 'tree-reference'
685
def __init__(self, file_id, name, parent_id, revision=None,
686
reference_revision=None):
687
InventoryEntry.__init__(self, file_id, name, parent_id)
688
self.revision = revision
689
self.reference_revision = reference_revision
692
return TreeReference(self.file_id, self.name, self.parent_id,
693
self.revision, self.reference_revision)
695
def _read_tree_state(self, path, work_tree):
696
"""Populate fields in the inventory entry from the given tree.
698
self.reference_revision = work_tree.get_reference_revision(
701
def _forget_tree_state(self):
702
self.reference_revision = None
704
def _unchanged(self, previous_ie):
705
"""See InventoryEntry._unchanged."""
706
compatible = super(TreeReference, self)._unchanged(previous_ie)
707
if self.reference_revision != previous_ie.reference_revision:
712
class Inventory(object):
199
713
"""Inventory of versioned files in a tree.
201
An Inventory acts like a set of InventoryEntry items. You can
202
also look files up by their file_id or name.
204
May be read from and written to a metadata file in a tree. To
205
manipulate the inventory (for example to add a file), it is read
206
in, modified, and then written back out.
715
This describes which file_id is present at each point in the tree,
716
and possibly the SHA-1 or other information about the file.
717
Entries can be looked up either by path or by file_id.
208
719
The inventory represents a typical unix file tree, with
209
720
directories containing files and subdirectories. We never store
236
742
May also look up by name:
238
744
>>> [x[0] for x in inv.iter_entries()]
241
>>> inv.write_xml(sys.stdout)
243
<entry file_id="123-123" kind="file" name="hello.c" />
746
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
747
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
748
Traceback (most recent call last):
749
BzrError: parent_id {TREE_ROOT} not in inventory
750
>>> inv.add(InventoryFile('123-123', 'hello.c', 'TREE_ROOT-12345678-12345678'))
751
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT-12345678-12345678', sha1=None, len=None)
248
## TODO: Clear up handling of files in subdirectories; we probably
249
## do want to be able to just look them up by name but this
250
## probably means gradually walking down the path, looking up as we go.
252
## TODO: Make sure only canonical filenames are stored.
254
## TODO: Do something sensible about the possible collisions on
255
## case-losing filesystems. Perhaps we should just always forbid
258
## _tree should probably just be stored as
259
## InventoryEntry._children on each directory.
753
def __init__(self, root_id=ROOT_ID, revision_id=None):
262
754
"""Create or read an inventory.
264
756
If a working directory is specified, the inventory is read
265
757
from there. If the file is specified, read from that. If not,
266
758
the inventory is created empty.
270
# _tree is indexed by parent_id; at each level a map from name
271
# to ie. The None entry is the root.
272
self._tree = {None: {}}
760
The inventory is created with a default root directory, with
763
if root_id is not None:
764
self._set_root(InventoryDirectory(root_id, u'', None))
768
self.revision_id = revision_id
771
return "<Inventory object at %x, contents=%r>" % (id(self), self._byid)
773
def apply_delta(self, delta):
774
"""Apply a delta to this inventory.
776
:param delta: A list of changes to apply. After all the changes are
777
applied the final inventory must be internally consistent, but it
778
is ok to supply changes which, if only half-applied would have an
779
invalid result - such as supplying two changes which rename two
780
files, 'A' and 'B' with each other : [('A', 'B', 'A-id', a_entry),
781
('B', 'A', 'B-id', b_entry)].
783
Each change is a tuple, of the form (old_path, new_path, file_id,
786
When new_path is None, the change indicates the removal of an entry
787
from the inventory and new_entry will be ignored (using None is
788
appropriate). If new_path is not None, then new_entry must be an
789
InventoryEntry instance, which will be incorporated into the
790
inventory (and replace any existing entry with the same file id).
792
When old_path is None, the change indicates the addition of
793
a new entry to the inventory.
795
When neither new_path nor old_path are None, the change is a
796
modification to an entry, such as a rename, reparent, kind change
799
The children attribute of new_entry is ignored. This is because
800
this method preserves children automatically across alterations to
801
the parent of the children, and cases where the parent id of a
802
child is changing require the child to be passed in as a separate
803
change regardless. E.g. in the recursive deletion of a directory -
804
the directory's children must be included in the delta, or the
805
final inventory will be invalid.
808
# Remove all affected items which were in the original inventory,
809
# starting with the longest paths, thus ensuring parents are examined
810
# after their children, which means that everything we examine has no
811
# modified children remaining by the time we examine it.
812
for old_path, file_id in sorted(((op, f) for op, np, f, e in delta
813
if op is not None), reverse=True):
814
if file_id not in self:
817
# Preserve unaltered children of file_id for later reinsertion.
818
file_id_children = getattr(self[file_id], 'children', {})
819
if len(file_id_children):
820
children[file_id] = file_id_children
821
# Remove file_id and the unaltered children. If file_id is not
822
# being deleted it will be reinserted back later.
823
self.remove_recursive_id(file_id)
824
# Insert all affected which should be in the new inventory, reattaching
825
# their children if they had any. This is done from shortest path to
826
# longest, ensuring that items which were modified and whose parents in
827
# the resulting inventory were also modified, are inserted after their
829
for new_path, new_entry in sorted((np, e) for op, np, f, e in
830
delta if np is not None):
831
if new_entry.kind == 'directory':
832
# Pop the child which to allow detection of children whose
833
# parents were deleted and which were not reattached to a new
835
new_entry.children = children.pop(new_entry.file_id, {})
838
# Get the parent id that was deleted
839
parent_id, children = children.popitem()
840
raise errors.InconsistentDelta("<deleted>", parent_id,
841
"The file id was deleted but its children were not deleted.")
843
def _set_root(self, ie):
845
self._byid = {self.root.file_id: self.root}
848
# TODO: jam 20051218 Should copy also copy the revision_id?
849
entries = self.iter_entries()
850
if self.root is None:
851
return Inventory(root_id=None)
852
other = Inventory(entries.next()[1].file_id)
853
other.root.revision = self.root.revision
854
# copy recursively so we know directories will be added before
855
# their children. There are more efficient ways than this...
856
for path, entry in entries:
857
other.add(entry.copy())
275
860
def __iter__(self):
276
861
return iter(self._byid)
279
863
def __len__(self):
280
864
"""Returns number of entries."""
281
865
return len(self._byid)
284
def iter_entries(self, parent_id=None):
867
def iter_entries(self, from_dir=None):
285
868
"""Return (path, entry) pairs, in order by name."""
286
kids = self._tree[parent_id].items()
288
for name, ie in kids:
290
if ie.kind == 'directory':
291
for cn, cie in self.iter_entries(parent_id=ie.file_id):
292
yield joinpath([name, cn]), cie
295
def directories(self, include_root=True):
296
"""Return (path, entry) pairs for all directories.
300
for path, entry in self.iter_entries():
301
if entry.kind == 'directory':
870
if self.root is None:
874
elif isinstance(from_dir, basestring):
875
from_dir = self._byid[from_dir]
877
# unrolling the recursive called changed the time from
878
# 440ms/663ms (inline/total) to 116ms/116ms
879
children = from_dir.children.items()
881
children = collections.deque(children)
882
stack = [(u'', children)]
884
from_dir_relpath, children = stack[-1]
887
name, ie = children.popleft()
889
# we know that from_dir_relpath never ends in a slash
890
# and 'f' doesn't begin with one, we can do a string op, rather
891
# than the checks of pathjoin(), though this means that all paths
893
path = from_dir_relpath + '/' + name
897
if ie.kind != 'directory':
900
# But do this child first
901
new_children = ie.children.items()
903
new_children = collections.deque(new_children)
904
stack.append((path, new_children))
905
# Break out of inner loop, so that we start outer loop with child
908
# if we finished all children, pop it off the stack
911
def iter_entries_by_dir(self, from_dir=None, specific_file_ids=None,
912
yield_parents=False):
913
"""Iterate over the entries in a directory first order.
915
This returns all entries for a directory before returning
916
the entries for children of a directory. This is not
917
lexicographically sorted order, and is a hybrid between
918
depth-first and breadth-first.
920
:param yield_parents: If True, yield the parents from the root leading
921
down to specific_file_ids that have been requested. This has no
922
impact if specific_file_ids is None.
923
:return: This yields (path, entry) pairs
925
if specific_file_ids and not isinstance(specific_file_ids, set):
926
specific_file_ids = set(specific_file_ids)
927
# TODO? Perhaps this should return the from_dir so that the root is
928
# yielded? or maybe an option?
930
if self.root is None:
932
# Optimize a common case
933
if (not yield_parents and specific_file_ids is not None and
934
len(specific_file_ids) == 1):
935
file_id = list(specific_file_ids)[0]
937
yield self.id2path(file_id), self[file_id]
940
if (specific_file_ids is None or yield_parents or
941
self.root.file_id in specific_file_ids):
943
elif isinstance(from_dir, basestring):
944
from_dir = self._byid[from_dir]
946
if specific_file_ids is not None:
947
# TODO: jam 20070302 This could really be done as a loop rather
948
# than a bunch of recursive calls.
951
def add_ancestors(file_id):
952
if file_id not in byid:
954
parent_id = byid[file_id].parent_id
955
if parent_id is None:
957
if parent_id not in parents:
958
parents.add(parent_id)
959
add_ancestors(parent_id)
960
for file_id in specific_file_ids:
961
add_ancestors(file_id)
965
stack = [(u'', from_dir)]
967
cur_relpath, cur_dir = stack.pop()
970
for child_name, child_ie in sorted(cur_dir.children.iteritems()):
972
child_relpath = cur_relpath + child_name
974
if (specific_file_ids is None or
975
child_ie.file_id in specific_file_ids or
976
(yield_parents and child_ie.file_id in parents)):
977
yield child_relpath, child_ie
979
if child_ie.kind == 'directory':
980
if parents is None or child_ie.file_id in parents:
981
child_dirs.append((child_relpath+'/', child_ie))
982
stack.extend(reversed(child_dirs))
984
def make_entry(self, kind, name, parent_id, file_id=None):
985
"""Simple thunk to bzrlib.inventory.make_entry."""
986
return make_entry(kind, name, parent_id, file_id)
989
"""Return list of (path, ie) for all entries except the root.
991
This may be faster than iter_entries.
994
def descend(dir_ie, dir_path):
995
kids = dir_ie.children.items()
997
for name, ie in kids:
998
child_path = osutils.pathjoin(dir_path, name)
999
accum.append((child_path, ie))
1000
if ie.kind == 'directory':
1001
descend(ie, child_path)
1003
descend(self.root, u'')
1006
def directories(self):
1007
"""Return (path, entry) pairs for all directories, including the root.
1010
def descend(parent_ie, parent_path):
1011
accum.append((parent_path, parent_ie))
1013
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
1016
for name, child_ie in kids:
1017
child_path = osutils.pathjoin(parent_path, name)
1018
descend(child_ie, child_path)
1019
descend(self.root, u'')
306
def children(self, parent_id):
307
"""Return entries that are direct children of parent_id."""
308
return self._tree[parent_id]
312
# TODO: return all paths and entries
315
1022
def __contains__(self, file_id):
316
1023
"""True if this entry contains a file with given id.
318
1025
>>> inv = Inventory()
319
>>> inv.add(InventoryEntry('123', 'foo.c'))
1026
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1027
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
320
1028
>>> '123' in inv
322
1030
>>> '456' in inv
325
return file_id in self._byid
1033
return (file_id in self._byid)
328
1035
def __getitem__(self, file_id):
329
1036
"""Return the entry for given file_id.
331
1038
>>> inv = Inventory()
332
>>> inv.add(InventoryEntry('123123', 'hello.c'))
1039
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
1040
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
333
1041
>>> inv['123123'].name
336
return self._byid[file_id]
1045
return self._byid[file_id]
1047
# really we're passing an inventory, not a tree...
1048
raise errors.NoSuchId(self, file_id)
1050
def get_file_kind(self, file_id):
1051
return self._byid[file_id].kind
1053
def get_child(self, parent_id, filename):
1054
return self[parent_id].children.get(filename)
1056
def _add_child(self, entry):
1057
"""Add an entry to the inventory, without adding it to its parent"""
1058
if entry.file_id in self._byid:
1059
raise BzrError("inventory already contains entry with id {%s}" %
1061
self._byid[entry.file_id] = entry
1062
for child in getattr(entry, 'children', {}).itervalues():
1063
self._add_child(child)
339
1066
def add(self, entry):
340
1067
"""Add entry to inventory.
342
1069
To add a file to a branch ready to be committed, use Branch.add,
344
if entry.file_id in self:
345
bailout("inventory already contains entry with id {%s}" % entry.file_id)
347
if entry.parent_id != None:
348
if entry.parent_id not in self:
349
bailout("parent_id %s of new entry not found in inventory"
352
if self._tree[entry.parent_id].has_key(entry.name):
353
bailout("%s is already versioned"
354
% appendpath(self.id2path(entry.parent_id), entry.name))
356
self._byid[entry.file_id] = entry
357
self._tree[entry.parent_id][entry.name] = entry
359
if entry.kind == 'directory':
360
self._tree[entry.file_id] = {}
1072
Returns the new entry object.
1074
if entry.file_id in self._byid:
1075
raise errors.DuplicateFileId(entry.file_id,
1076
self._byid[entry.file_id])
1078
if entry.parent_id is None:
1082
parent = self._byid[entry.parent_id]
1084
raise BzrError("parent_id {%s} not in inventory" %
1087
if entry.name in parent.children:
1088
raise BzrError("%s is already versioned" %
1089
osutils.pathjoin(self.id2path(parent.file_id),
1090
entry.name).encode('utf-8'))
1091
parent.children[entry.name] = entry
1092
return self._add_child(entry)
1094
def add_path(self, relpath, kind, file_id=None, parent_id=None):
1095
"""Add entry from a path.
1097
The immediate parent must already be versioned.
1099
Returns the new entry object."""
1101
parts = osutils.splitpath(relpath)
1105
file_id = generate_ids.gen_root_id()
1106
self.root = InventoryDirectory(file_id, '', None)
1107
self._byid = {self.root.file_id: self.root}
1110
parent_path = parts[:-1]
1111
parent_id = self.path2id(parent_path)
1112
if parent_id is None:
1113
raise errors.NotVersionedError(path=parent_path)
1114
ie = make_entry(kind, parts[-1], parent_id, file_id)
363
1117
def __delitem__(self, file_id):
364
1118
"""Remove entry by id.
366
1120
>>> inv = Inventory()
367
>>> inv.add(InventoryEntry('123', 'foo.c'))
1121
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1122
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
368
1123
>>> '123' in inv
370
1125
>>> del inv['123']
added
removed
bzrlib/inventory.py
