1
# (C) 2005 Canonical Ltd
1
# Copyright (C) 2005, 2006 Canonical Ltd
3
3
# This program is free software; you can redistribute it and/or modify
4
4
# it under the terms of the GNU General Public License as published by
5
5
# the Free Software Foundation; either version 2 of the License, or
6
6
# (at your option) any later version.
8
8
# This program is distributed in the hope that it will be useful,
9
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
11
# GNU General Public License for more details.
13
13
# You should have received a copy of the GNU General Public License
14
14
# along with this program; if not, write to the Free Software
15
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
# FIXME: This refactoring of the workingtree code doesn't seem to keep
18
# the WorkingTree's copy of the inventory in sync with the branch. The
19
# branch modifies its working inventory when it does a commit to make
20
# missing files permanently removed.
22
# TODO: Maybe also keep the full path of the entry, and the children?
23
# But those depend on its position within a particular inventory, and
24
# it would be nice not to need to hold the backpointer here.
18
26
# This should really be an id randomly assigned when the tree is
19
27
# created, but it's not for now.
20
28
ROOT_ID = "TREE_ROOT"
23
import sys, os.path, types, re
27
from cElementTree import Element, ElementTree, SubElement
29
from elementtree.ElementTree import Element, ElementTree, SubElement
31
from xml import XMLMixin
32
from errors import bailout, BzrError, BzrCheckError
34
from bzrlib.lazy_import import lazy_import
35
lazy_import(globals(), """
35
from bzrlib.osutils import uuid, quotefn, splitpath, joinpath, appendpath
49
from bzrlib.errors import (
36
53
from bzrlib.trace import mutter
38
class InventoryEntry(XMLMixin):
56
class InventoryEntry(object):
39
57
"""Description of a versioned file.
41
59
An InventoryEntry has the following fields, which are also
42
60
present in the XML inventory-entry element:
45
* *name*: (only the basename within the directory, must not
47
* *kind*: "directory" or "file"
48
* *directory_id*: (if absent/null means the branch root directory)
49
* *text_sha1*: only for files
50
* *text_size*: in bytes, only for files
51
* *text_id*: identifier for the text version, only for files
53
InventoryEntries can also exist inside a WorkingTree
54
inventory, in which case they are not yet bound to a
55
particular revision of the file. In that case the text_sha1,
56
text_size and text_id are absent.
65
(within the parent directory)
68
file_id of the parent directory, or ROOT_ID
71
the revision_id in which this variation of this file was
75
Indicates that this file should be executable on systems
79
sha-1 of the text of the file
82
size in bytes of the text of the file
84
(reading a version 4 tree created a text_id field.)
59
86
>>> i = Inventory()
62
>>> i.add(InventoryEntry('123', 'src', 'directory', ROOT_ID))
63
>>> i.add(InventoryEntry('2323', 'hello.c', 'file', parent_id='123'))
64
>>> for j in i.iter_entries():
89
>>> i.add(InventoryDirectory('123', 'src', ROOT_ID))
90
InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None)
91
>>> i.add(InventoryFile('2323', 'hello.c', parent_id='123'))
92
InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None)
93
>>> shouldbe = {0: '', 1: 'src', 2: 'src/hello.c'}
94
>>> for ix, j in enumerate(i.iter_entries()):
95
... print (j[0] == shouldbe[ix], j[1])
67
('src', InventoryEntry('123', 'src', kind='directory', parent_id='TREE_ROOT'))
68
('src/hello.c', InventoryEntry('2323', 'hello.c', kind='file', parent_id='123'))
69
>>> i.add(InventoryEntry('2323', 'bye.c', 'file', '123'))
97
(True, InventoryDirectory('TREE_ROOT', u'', parent_id=None, revision=None))
98
(True, InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None))
99
(True, InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None))
100
>>> i.add(InventoryFile('2323', 'bye.c', '123'))
70
101
Traceback (most recent call last):
72
BzrError: ('inventory already contains entry with id {2323}', [])
73
>>> i.add(InventoryEntry('2324', 'bye.c', 'file', '123'))
74
>>> i.add(InventoryEntry('2325', 'wibble', 'directory', '123'))
103
BzrError: inventory already contains entry with id {2323}
104
>>> i.add(InventoryFile('2324', 'bye.c', '123'))
105
InventoryFile('2324', 'bye.c', parent_id='123', sha1=None, len=None)
106
>>> i.add(InventoryDirectory('2325', 'wibble', '123'))
107
InventoryDirectory('2325', 'wibble', parent_id='123', revision=None)
75
108
>>> i.path2id('src/wibble')
79
>>> i.add(InventoryEntry('2326', 'wibble.c', 'file', '2325'))
112
>>> i.add(InventoryFile('2326', 'wibble.c', '2325'))
113
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
81
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
82
>>> for j in i.iter_entries():
84
... assert i.path2id(j[0])
115
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
116
>>> for path, entry in i.iter_entries():
118
... assert i.path2id(path)
90
125
src/wibble/wibble.c
91
126
>>> i.id2path('2326')
92
127
'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.
99
# TODO: split InventoryEntry into subclasses for files,
100
# directories, etc etc.
130
# Constants returned by describe_change()
132
# TODO: These should probably move to some kind of FileChangeDescription
133
# class; that's like what's inside a TreeDelta but we want to be able to
134
# generate them just for one file at a time.
136
MODIFIED_AND_RENAMED = 'modified and renamed'
105
def __init__(self, file_id, name, kind, parent_id, text_id=None):
140
def detect_changes(self, old_entry):
141
"""Return a (text_modified, meta_modified) from this to old_entry.
143
_read_tree_state must have been called on self and old_entry prior to
144
calling detect_changes.
148
def diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
149
output_to, reverse=False):
150
"""Perform a diff from this to to_entry.
152
text_diff will be used for textual difference calculation.
153
This is a template method, override _diff in child classes.
155
self._read_tree_state(tree.id2path(self.file_id), tree)
157
# cannot diff from one kind to another - you must do a removal
158
# and an addif they do not match.
159
assert self.kind == to_entry.kind
160
to_entry._read_tree_state(to_tree.id2path(to_entry.file_id),
162
self._diff(text_diff, from_label, tree, to_label, to_entry, to_tree,
165
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
166
output_to, reverse=False):
167
"""Perform a diff between two entries of the same kind."""
169
def find_previous_heads(self, previous_inventories,
170
versioned_file_store,
173
"""Return the revisions and entries that directly precede this.
175
Returned as a map from revision to inventory entry.
177
This is a map containing the file revisions in all parents
178
for which the file exists, and its revision is not a parent of
179
any other. If the file is new, the set will be empty.
181
:param versioned_file_store: A store where ancestry data on this
182
file id can be queried.
183
:param transaction: The transaction that queries to the versioned
184
file store should be completed under.
185
:param entry_vf: The entry versioned file, if its already available.
187
def get_ancestors(weave, entry):
188
return set(weave.get_ancestry(entry.revision))
189
# revision:ie mapping for each ie found in previous_inventories.
191
# revision:ie mapping with one revision for each head.
193
# revision: ancestor list for each head
195
# identify candidate head revision ids.
196
for inv in previous_inventories:
197
if self.file_id in inv:
198
ie = inv[self.file_id]
199
assert ie.file_id == self.file_id
200
if ie.revision in candidates:
201
# same revision value in two different inventories:
202
# correct possible inconsistencies:
203
# * there was a bug in revision updates with 'x' bit
206
if candidates[ie.revision].executable != ie.executable:
207
candidates[ie.revision].executable = False
208
ie.executable = False
209
except AttributeError:
211
# must now be the same.
212
assert candidates[ie.revision] == ie
214
# add this revision as a candidate.
215
candidates[ie.revision] = ie
217
# common case optimisation
218
if len(candidates) == 1:
219
# if there is only one candidate revision found
220
# then we can opening the versioned file to access ancestry:
221
# there cannot be any ancestors to eliminate when there is
222
# only one revision available.
223
heads[ie.revision] = ie
226
# eliminate ancestors amongst the available candidates:
227
# heads are those that are not an ancestor of any other candidate
228
# - this provides convergence at a per-file level.
229
for ie in candidates.values():
230
# may be an ancestor of a known head:
231
already_present = 0 != len(
232
[head for head in heads
233
if ie.revision in head_ancestors[head]])
235
# an ancestor of an analyzed candidate.
237
# not an ancestor of a known head:
238
# load the versioned file for this file id if needed
240
entry_vf = versioned_file_store.get_weave_or_empty(
241
self.file_id, transaction)
242
ancestors = get_ancestors(entry_vf, ie)
243
# may knock something else out:
244
check_heads = list(heads.keys())
245
for head in check_heads:
246
if head in ancestors:
247
# this previously discovered 'head' is not
248
# really a head - its an ancestor of the newly
251
head_ancestors[ie.revision] = ancestors
252
heads[ie.revision] = ie
255
def get_tar_item(self, root, dp, now, tree):
256
"""Get a tarfile item and a file stream for its content."""
257
item = tarfile.TarInfo(osutils.pathjoin(root, dp).encode('utf8'))
258
# TODO: would be cool to actually set it to the timestamp of the
259
# revision it was last changed
261
fileobj = self._put_in_tar(item, tree)
265
"""Return true if the object this entry represents has textual data.
267
Note that textual data includes binary content.
269
Also note that all entries get weave files created for them.
270
This attribute is primarily used when upgrading from old trees that
271
did not have the weave index for all inventory entries.
275
def __init__(self, file_id, name, parent_id, text_id=None):
106
276
"""Create an InventoryEntry
108
278
The filename must be a single component, relative to the
109
279
parent directory; it cannot be a whole path or relative name.
111
>>> e = InventoryEntry('123', 'hello.c', 'file', ROOT_ID)
281
>>> e = InventoryFile('123', 'hello.c', ROOT_ID)
116
>>> e = InventoryEntry('123', 'src/hello.c', 'file', ROOT_ID)
286
>>> e = InventoryFile('123', 'src/hello.c', ROOT_ID)
117
287
Traceback (most recent call last):
118
BzrCheckError: InventoryEntry name 'src/hello.c' is invalid
288
InvalidEntryName: Invalid entry name: src/hello.c
290
assert isinstance(name, basestring), name
120
291
if '/' in name or '\\' in name:
121
raise BzrCheckError('InventoryEntry name %r is invalid' % name)
292
raise errors.InvalidEntryName(name=name)
293
self.executable = False
295
self.text_sha1 = None
296
self.text_size = None
123
297
self.file_id = file_id
298
assert isinstance(file_id, (str, None.__class__)), \
299
'bad type %r for %r' % (type(file_id), file_id)
126
301
self.text_id = text_id
127
302
self.parent_id = parent_id
128
if kind == 'directory':
133
raise BzrError("unhandled entry kind %r" % kind)
303
self.symlink_target = None
305
def kind_character(self):
306
"""Return a short kind indicator useful for appending to names."""
307
raise BzrError('unknown kind %r' % self.kind)
309
known_kinds = ('file', 'directory', 'symlink')
311
def _put_in_tar(self, item, tree):
312
"""populate item for stashing in a tar, and return the content stream.
314
If no content is available, return None.
316
raise BzrError("don't know how to export {%s} of kind %r" %
317
(self.file_id, self.kind))
319
def put_on_disk(self, dest, dp, tree):
320
"""Create a representation of self on disk in the prefix dest.
322
This is a template method - implement _put_on_disk in subclasses.
324
fullpath = osutils.pathjoin(dest, dp)
325
self._put_on_disk(fullpath, tree)
326
# mutter(" export {%s} kind %s to %s", self.file_id,
327
# self.kind, fullpath)
329
def _put_on_disk(self, fullpath, tree):
330
"""Put this entry onto disk at fullpath, from tree tree."""
331
raise BzrError("don't know how to export {%s} of kind %r" % (self.file_id, self.kind))
137
333
def sorted_children(self):
138
l = self.children.items()
334
return sorted(self.children.items())
337
def versionable_kind(kind):
338
return (kind in ('file', 'directory', 'symlink'))
340
def check(self, checker, rev_id, inv, tree):
341
"""Check this inventory entry is intact.
343
This is a template method, override _check for kind specific
346
:param checker: Check object providing context for the checks;
347
can be used to find out what parts of the repository have already
349
:param rev_id: Revision id from which this InventoryEntry was loaded.
350
Not necessarily the last-changed revision for this file.
351
:param inv: Inventory from which the entry was loaded.
352
:param tree: RevisionTree for this entry.
354
if self.parent_id is not None:
355
if not inv.has_id(self.parent_id):
356
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
357
% (self.parent_id, rev_id))
358
self._check(checker, rev_id, tree)
360
def _check(self, checker, rev_id, tree):
361
"""Check this inventory entry for kind specific errors."""
362
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
144
other = InventoryEntry(self.file_id, self.name, self.kind,
145
self.parent_id, text_id=self.text_id)
146
other.text_sha1 = self.text_sha1
147
other.text_size = self.text_size
366
"""Clone this inventory entry."""
367
raise NotImplementedError
370
def describe_change(old_entry, new_entry):
371
"""Describe the change between old_entry and this.
373
This smells of being an InterInventoryEntry situation, but as its
374
the first one, we're making it a static method for now.
376
An entry with a different parent, or different name is considered
377
to be renamed. Reparenting is an internal detail.
378
Note that renaming the parent does not trigger a rename for the
381
# TODO: Perhaps return an object rather than just a string
382
if old_entry is new_entry:
383
# also the case of both being None
385
elif old_entry is None:
387
elif new_entry is None:
389
if old_entry.kind != new_entry.kind:
391
text_modified, meta_modified = new_entry.detect_changes(old_entry)
392
if text_modified or meta_modified:
396
# TODO 20060511 (mbp, rbc) factor out 'detect_rename' here.
397
if old_entry.parent_id != new_entry.parent_id:
399
elif old_entry.name != new_entry.name:
403
if renamed and not modified:
404
return InventoryEntry.RENAMED
405
if modified and not renamed:
407
if modified and renamed:
408
return InventoryEntry.MODIFIED_AND_RENAMED
151
411
def __repr__(self):
152
return ("%s(%r, %r, kind=%r, parent_id=%r)"
412
return ("%s(%r, %r, parent_id=%r, revision=%r)"
153
413
% (self.__class__.__name__,
160
def to_element(self):
161
"""Convert to XML element"""
164
e.set('name', self.name)
165
e.set('file_id', self.file_id)
166
e.set('kind', self.kind)
168
if self.text_size != None:
169
e.set('text_size', '%d' % self.text_size)
171
for f in ['text_id', 'text_sha1']:
176
# to be conservative, we don't externalize the root pointers
177
# for now, leaving them as null in the xml form. in a future
178
# version it will be implied by nested elements.
179
if self.parent_id != ROOT_ID:
180
assert isinstance(self.parent_id, basestring)
181
e.set('parent_id', self.parent_id)
188
def from_element(cls, elt):
189
assert elt.tag == 'entry'
191
## original format inventories don't have a parent_id for
192
## nodes in the root directory, but it's cleaner to use one
194
parent_id = elt.get('parent_id')
195
if parent_id == None:
198
self = cls(elt.get('file_id'), elt.get('name'), elt.get('kind'), parent_id)
199
self.text_id = elt.get('text_id')
200
self.text_sha1 = elt.get('text_sha1')
202
## mutter("read inventoryentry: %r" % (elt.attrib))
204
v = elt.get('text_size')
205
self.text_size = v and int(v)
210
from_element = classmethod(from_element)
212
def __cmp__(self, other):
419
def snapshot(self, revision, path, previous_entries,
420
work_tree, commit_builder):
421
"""Make a snapshot of this entry which may or may not have changed.
423
This means that all its fields are populated, that it has its
424
text stored in the text store or weave.
426
# mutter('new parents of %s are %r', path, previous_entries)
427
self._read_tree_state(path, work_tree)
428
# TODO: Where should we determine whether to reuse a
429
# previous revision id or create a new revision? 20060606
430
if len(previous_entries) == 1:
431
# cannot be unchanged unless there is only one parent file rev.
432
parent_ie = previous_entries.values()[0]
433
if self._unchanged(parent_ie):
434
# mutter("found unchanged entry")
435
self.revision = parent_ie.revision
437
return self._snapshot_into_revision(revision, previous_entries,
438
work_tree, commit_builder)
440
def _snapshot_into_revision(self, revision, previous_entries, work_tree,
442
"""Record this revision unconditionally into a store.
444
The entry's last-changed revision property (`revision`) is updated to
445
that of the new revision.
447
:param revision: id of the new revision that is being recorded.
449
:returns: String description of the commit (e.g. "merged", "modified"), etc.
451
# mutter('new revision {%s} for {%s}', revision, self.file_id)
452
self.revision = revision
453
self._snapshot_text(previous_entries, work_tree, commit_builder)
455
def _snapshot_text(self, file_parents, work_tree, commit_builder):
456
"""Record the 'text' of this entry, whatever form that takes.
458
This default implementation simply adds an empty text.
460
raise NotImplementedError(self._snapshot_text)
462
def __eq__(self, other):
215
463
if not isinstance(other, InventoryEntry):
216
464
return NotImplemented
218
return cmp(self.file_id, other.file_id) \
219
or cmp(self.name, other.name) \
220
or cmp(self.text_sha1, other.text_sha1) \
221
or cmp(self.text_size, other.text_size) \
222
or cmp(self.text_id, other.text_id) \
223
or cmp(self.parent_id, other.parent_id) \
224
or cmp(self.kind, other.kind)
466
return ((self.file_id == other.file_id)
467
and (self.name == other.name)
468
and (other.symlink_target == self.symlink_target)
469
and (self.text_sha1 == other.text_sha1)
470
and (self.text_size == other.text_size)
471
and (self.text_id == other.text_id)
472
and (self.parent_id == other.parent_id)
473
and (self.kind == other.kind)
474
and (self.revision == other.revision)
475
and (self.executable == other.executable)
478
def __ne__(self, other):
479
return not (self == other)
482
raise ValueError('not hashable')
484
def _unchanged(self, previous_ie):
485
"""Has this entry changed relative to previous_ie.
487
This method should be overridden in child classes.
490
# different inv parent
491
if previous_ie.parent_id != self.parent_id:
494
elif previous_ie.name != self.name:
498
def _read_tree_state(self, path, work_tree):
499
"""Populate fields in the inventory entry from the given tree.
501
Note that this should be modified to be a noop on virtual trees
502
as all entries created there are prepopulated.
504
# TODO: Rather than running this manually, we should check the
505
# working sha1 and other expensive properties when they're
506
# first requested, or preload them if they're already known
507
pass # nothing to do by default
509
def _forget_tree_state(self):
228
513
class RootEntry(InventoryEntry):
515
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
516
'text_id', 'parent_id', 'children', 'executable',
517
'revision', 'symlink_target']
519
def _check(self, checker, rev_id, tree):
520
"""See InventoryEntry._check"""
229
522
def __init__(self, file_id):
230
523
self.file_id = file_id
231
524
self.children = {}
232
self.kind = 'root_directory'
525
self.kind = 'directory'
233
526
self.parent_id = None
529
symbol_versioning.warn('RootEntry is deprecated as of bzr 0.10.'
530
' Please use InventoryDirectory instead.',
531
DeprecationWarning, stacklevel=2)
236
def __cmp__(self, other):
533
def __eq__(self, other):
239
534
if not isinstance(other, RootEntry):
240
535
return NotImplemented
241
return cmp(self.file_id, other.file_id) \
242
or cmp(self.children, other.children)
246
class Inventory(XMLMixin):
537
return (self.file_id == other.file_id) \
538
and (self.children == other.children)
541
class InventoryDirectory(InventoryEntry):
542
"""A directory in an inventory."""
544
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
545
'text_id', 'parent_id', 'children', 'executable',
546
'revision', 'symlink_target']
548
def _check(self, checker, rev_id, tree):
549
"""See InventoryEntry._check"""
550
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
551
raise BzrCheckError('directory {%s} has text in revision {%s}'
552
% (self.file_id, rev_id))
555
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
556
other.revision = self.revision
557
# note that children are *not* copied; they're pulled across when
561
def __init__(self, file_id, name, parent_id):
562
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
564
self.kind = 'directory'
566
def kind_character(self):
567
"""See InventoryEntry.kind_character."""
570
def _put_in_tar(self, item, tree):
571
"""See InventoryEntry._put_in_tar."""
572
item.type = tarfile.DIRTYPE
579
def _put_on_disk(self, fullpath, tree):
580
"""See InventoryEntry._put_on_disk."""
583
def _snapshot_text(self, file_parents, work_tree, commit_builder):
584
"""See InventoryEntry._snapshot_text."""
585
commit_builder.modified_directory(self.file_id, file_parents)
588
class InventoryFile(InventoryEntry):
589
"""A file in an inventory."""
591
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
592
'text_id', 'parent_id', 'children', 'executable',
593
'revision', 'symlink_target']
595
def _check(self, checker, tree_revision_id, tree):
596
"""See InventoryEntry._check"""
597
t = (self.file_id, self.revision)
598
if t in checker.checked_texts:
599
prev_sha = checker.checked_texts[t]
600
if prev_sha != self.text_sha1:
601
raise BzrCheckError('mismatched sha1 on {%s} in {%s}' %
602
(self.file_id, tree_revision_id))
604
checker.repeated_text_cnt += 1
607
if self.file_id not in checker.checked_weaves:
608
mutter('check weave {%s}', self.file_id)
609
w = tree.get_weave(self.file_id)
610
# Not passing a progress bar, because it creates a new
611
# progress, which overwrites the current progress,
612
# and doesn't look nice
614
checker.checked_weaves[self.file_id] = True
616
w = tree.get_weave(self.file_id)
618
mutter('check version {%s} of {%s}', tree_revision_id, self.file_id)
619
checker.checked_text_cnt += 1
620
# We can't check the length, because Weave doesn't store that
621
# information, and the whole point of looking at the weave's
622
# sha1sum is that we don't have to extract the text.
623
if self.text_sha1 != w.get_sha1(self.revision):
624
raise BzrCheckError('text {%s} version {%s} wrong sha1'
625
% (self.file_id, self.revision))
626
checker.checked_texts[t] = self.text_sha1
629
other = InventoryFile(self.file_id, self.name, self.parent_id)
630
other.executable = self.executable
631
other.text_id = self.text_id
632
other.text_sha1 = self.text_sha1
633
other.text_size = self.text_size
634
other.revision = self.revision
637
def detect_changes(self, old_entry):
638
"""See InventoryEntry.detect_changes."""
639
assert self.text_sha1 is not None
640
assert old_entry.text_sha1 is not None
641
text_modified = (self.text_sha1 != old_entry.text_sha1)
642
meta_modified = (self.executable != old_entry.executable)
643
return text_modified, meta_modified
645
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
646
output_to, reverse=False):
647
"""See InventoryEntry._diff."""
649
from_text = tree.get_file(self.file_id).readlines()
651
to_text = to_tree.get_file(to_entry.file_id).readlines()
655
text_diff(from_label, from_text,
656
to_label, to_text, output_to)
658
text_diff(to_label, to_text,
659
from_label, from_text, output_to)
660
except errors.BinaryFile:
662
label_pair = (to_label, from_label)
664
label_pair = (from_label, to_label)
665
print >> output_to, "Binary files %s and %s differ" % label_pair
668
"""See InventoryEntry.has_text."""
671
def __init__(self, file_id, name, parent_id):
672
super(InventoryFile, self).__init__(file_id, name, parent_id)
675
def kind_character(self):
676
"""See InventoryEntry.kind_character."""
679
def _put_in_tar(self, item, tree):
680
"""See InventoryEntry._put_in_tar."""
681
item.type = tarfile.REGTYPE
682
fileobj = tree.get_file(self.file_id)
683
item.size = self.text_size
684
if tree.is_executable(self.file_id):
690
def _put_on_disk(self, fullpath, tree):
691
"""See InventoryEntry._put_on_disk."""
692
osutils.pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
693
if tree.is_executable(self.file_id):
694
os.chmod(fullpath, 0755)
696
def _read_tree_state(self, path, work_tree):
697
"""See InventoryEntry._read_tree_state."""
698
self.text_sha1 = work_tree.get_file_sha1(self.file_id, path=path)
699
# FIXME: 20050930 probe for the text size when getting sha1
700
# in _read_tree_state
701
self.executable = work_tree.is_executable(self.file_id, path=path)
704
return ("%s(%r, %r, parent_id=%r, sha1=%r, len=%s)"
705
% (self.__class__.__name__,
712
def _forget_tree_state(self):
713
self.text_sha1 = None
715
def _snapshot_text(self, file_parents, work_tree, commit_builder):
716
"""See InventoryEntry._snapshot_text."""
717
def get_content_byte_lines():
718
return work_tree.get_file(self.file_id).readlines()
719
self.text_sha1, self.text_size = commit_builder.modified_file_text(
720
self.file_id, file_parents, get_content_byte_lines, self.text_sha1, self.text_size)
722
def _unchanged(self, previous_ie):
723
"""See InventoryEntry._unchanged."""
724
compatible = super(InventoryFile, self)._unchanged(previous_ie)
725
if self.text_sha1 != previous_ie.text_sha1:
728
# FIXME: 20050930 probe for the text size when getting sha1
729
# in _read_tree_state
730
self.text_size = previous_ie.text_size
731
if self.executable != previous_ie.executable:
736
class InventoryLink(InventoryEntry):
737
"""A file in an inventory."""
739
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
740
'text_id', 'parent_id', 'children', 'executable',
741
'revision', 'symlink_target']
743
def _check(self, checker, rev_id, tree):
744
"""See InventoryEntry._check"""
745
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
746
raise BzrCheckError('symlink {%s} has text in revision {%s}'
747
% (self.file_id, rev_id))
748
if self.symlink_target is None:
749
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
750
% (self.file_id, rev_id))
753
other = InventoryLink(self.file_id, self.name, self.parent_id)
754
other.symlink_target = self.symlink_target
755
other.revision = self.revision
758
def detect_changes(self, old_entry):
759
"""See InventoryEntry.detect_changes."""
760
# FIXME: which _modified field should we use ? RBC 20051003
761
text_modified = (self.symlink_target != old_entry.symlink_target)
763
mutter(" symlink target changed")
764
meta_modified = False
765
return text_modified, meta_modified
767
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
768
output_to, reverse=False):
769
"""See InventoryEntry._diff."""
770
from_text = self.symlink_target
771
if to_entry is not None:
772
to_text = to_entry.symlink_target
777
print >>output_to, '=== target changed %r => %r' % (from_text, to_text)
780
print >>output_to, '=== target was %r' % self.symlink_target
782
print >>output_to, '=== target is %r' % self.symlink_target
784
def __init__(self, file_id, name, parent_id):
785
super(InventoryLink, self).__init__(file_id, name, parent_id)
786
self.kind = 'symlink'
788
def kind_character(self):
789
"""See InventoryEntry.kind_character."""
792
def _put_in_tar(self, item, tree):
793
"""See InventoryEntry._put_in_tar."""
794
item.type = tarfile.SYMTYPE
798
item.linkname = self.symlink_target
801
def _put_on_disk(self, fullpath, tree):
802
"""See InventoryEntry._put_on_disk."""
804
os.symlink(self.symlink_target, fullpath)
806
raise BzrError("Failed to create symlink %r -> %r, error: %s" % (fullpath, self.symlink_target, e))
808
def _read_tree_state(self, path, work_tree):
809
"""See InventoryEntry._read_tree_state."""
810
self.symlink_target = work_tree.get_symlink_target(self.file_id)
812
def _forget_tree_state(self):
813
self.symlink_target = None
815
def _unchanged(self, previous_ie):
816
"""See InventoryEntry._unchanged."""
817
compatible = super(InventoryLink, self)._unchanged(previous_ie)
818
if self.symlink_target != previous_ie.symlink_target:
822
def _snapshot_text(self, file_parents, work_tree, commit_builder):
823
"""See InventoryEntry._snapshot_text."""
824
commit_builder.modified_link(
825
self.file_id, file_parents, self.symlink_target)
828
class Inventory(object):
247
829
"""Inventory of versioned files in a tree.
249
831
This describes which file_id is present at each point in the tree,
296
876
The inventory is created with a default root directory, with
299
self.root = RootEntry(ROOT_ID)
879
if root_id is not None:
880
assert root_id.__class__ == str
881
self._set_root(InventoryDirectory(root_id, u'', None))
885
self.revision_id = revision_id
887
def _set_root(self, ie):
300
889
self._byid = {self.root.file_id: self.root}
892
# TODO: jam 20051218 Should copy also copy the revision_id?
893
entries = self.iter_entries()
894
other = Inventory(entries.next()[1].file_id)
895
# copy recursively so we know directories will be added before
896
# their children. There are more efficient ways than this...
897
for path, entry in entries():
898
other.add(entry.copy())
303
901
def __iter__(self):
304
902
return iter(self._byid)
307
904
def __len__(self):
308
905
"""Returns number of entries."""
309
906
return len(self._byid)
312
908
def iter_entries(self, from_dir=None):
313
909
"""Return (path, entry) pairs, in order by name."""
317
elif isinstance(from_dir, basestring):
318
from_dir = self._byid[from_dir]
320
kids = from_dir.children.items()
322
for name, ie in kids:
324
if ie.kind == 'directory':
325
for cn, cie in self.iter_entries(from_dir=ie.file_id):
326
yield os.path.join(name, cn), cie
911
if self.root is None:
915
elif isinstance(from_dir, basestring):
916
from_dir = self._byid[from_dir]
918
# unrolling the recursive called changed the time from
919
# 440ms/663ms (inline/total) to 116ms/116ms
920
children = from_dir.children.items()
922
children = collections.deque(children)
923
stack = [(u'', children)]
925
from_dir_relpath, children = stack[-1]
928
name, ie = children.popleft()
930
# we know that from_dir_relpath never ends in a slash
931
# and 'f' doesn't begin with one, we can do a string op, rather
932
# than the checks of pathjoin(), though this means that all paths
934
path = from_dir_relpath + '/' + name
938
if ie.kind != 'directory':
941
# But do this child first
942
new_children = ie.children.items()
944
new_children = collections.deque(new_children)
945
stack.append((path, new_children))
946
# Break out of inner loop, so that we start outer loop with child
949
# if we finished all children, pop it off the stack
952
def iter_entries_by_dir(self, from_dir=None, specific_file_ids=None):
953
"""Iterate over the entries in a directory first order.
955
This returns all entries for a directory before returning
956
the entries for children of a directory. This is not
957
lexicographically sorted order, and is a hybrid between
958
depth-first and breadth-first.
960
:return: This yields (path, entry) pairs
962
if specific_file_ids:
963
specific_file_ids = [osutils.safe_file_id(fid)
964
for fid in specific_file_ids]
965
# TODO? Perhaps this should return the from_dir so that the root is
966
# yielded? or maybe an option?
968
if self.root is None:
970
# Optimize a common case
971
if specific_file_ids is not None and len(specific_file_ids) == 1:
972
file_id = list(specific_file_ids)[0]
974
yield self.id2path(file_id), self[file_id]
977
if (specific_file_ids is None or
978
self.root.file_id in specific_file_ids):
980
elif isinstance(from_dir, basestring):
981
from_dir = self._byid[from_dir]
983
if specific_file_ids is not None:
985
def add_ancestors(file_id):
986
if file_id not in self:
988
parent_id = self[file_id].parent_id
989
if parent_id is None:
991
if parent_id not in parents:
992
parents.add(parent_id)
993
add_ancestors(parent_id)
994
for file_id in specific_file_ids:
995
add_ancestors(file_id)
999
stack = [(u'', from_dir)]
1001
cur_relpath, cur_dir = stack.pop()
1004
for child_name, child_ie in sorted(cur_dir.children.iteritems()):
1006
child_relpath = cur_relpath + child_name
1008
if (specific_file_ids is None or
1009
child_ie.file_id in specific_file_ids):
1010
yield child_relpath, child_ie
1012
if child_ie.kind == 'directory':
1013
if parents is None or child_ie.file_id in parents:
1014
child_dirs.append((child_relpath+'/', child_ie))
1015
stack.extend(reversed(child_dirs))
1018
"""Return list of (path, ie) for all entries except the root.
1020
This may be faster than iter_entries.
1023
def descend(dir_ie, dir_path):
1024
kids = dir_ie.children.items()
1026
for name, ie in kids:
1027
child_path = osutils.pathjoin(dir_path, name)
1028
accum.append((child_path, ie))
1029
if ie.kind == 'directory':
1030
descend(ie, child_path)
1032
descend(self.root, u'')
330
1035
def directories(self):
331
"""Return (path, entry) pairs for all directories.
1036
"""Return (path, entry) pairs for all directories, including the root.
333
def descend(parent_ie):
334
parent_name = parent_ie.name
335
yield parent_name, parent_ie
337
# directory children in sorted order
339
for ie in parent_ie.children.itervalues():
340
if ie.kind == 'directory':
341
dn.append((ie.name, ie))
1039
def descend(parent_ie, parent_path):
1040
accum.append((parent_path, parent_ie))
344
for name, child_ie in dn:
345
for sub_name, sub_ie in descend(child_ie):
346
yield appendpath(parent_name, sub_name), sub_ie
1042
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
348
for name, ie in descend(self.root):
1045
for name, child_ie in kids:
1046
child_path = osutils.pathjoin(parent_path, name)
1047
descend(child_ie, child_path)
1048
descend(self.root, u'')
353
1051
def __contains__(self, file_id):
354
1052
"""True if this entry contains a file with given id.
356
1054
>>> inv = Inventory()
357
>>> inv.add(InventoryEntry('123', 'foo.c', 'file', ROOT_ID))
1055
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1056
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
358
1057
>>> '123' in inv
360
1059
>>> '456' in inv
363
return file_id in self._byid
1062
file_id = osutils.safe_file_id(file_id)
1063
return (file_id in self._byid)
366
1065
def __getitem__(self, file_id):
367
1066
"""Return the entry for given file_id.
369
1068
>>> inv = Inventory()
370
>>> inv.add(InventoryEntry('123123', 'hello.c', 'file', ROOT_ID))
1069
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
1070
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
371
1071
>>> inv['123123'].name
1074
file_id = osutils.safe_file_id(file_id)
375
1076
return self._byid[file_id]
376
1077
except KeyError:
378
raise BzrError("can't look up file_id None")
380
raise BzrError("file_id {%s} not in inventory" % file_id)
1078
# really we're passing an inventory, not a tree...
1079
raise errors.NoSuchId(self, file_id)
1081
def get_file_kind(self, file_id):
1082
file_id = osutils.safe_file_id(file_id)
1083
return self._byid[file_id].kind
383
1085
def get_child(self, parent_id, filename):
1086
parent_id = osutils.safe_file_id(parent_id)
384
1087
return self[parent_id].children.get(filename)
387
1089
def add(self, entry):
388
1090
"""Add entry to inventory.
390
1092
To add a file to a branch ready to be committed, use Branch.add,
1095
Returns the new entry object.
392
1097
if entry.file_id in self._byid:
393
bailout("inventory already contains entry with id {%s}" % entry.file_id)
1098
raise errors.DuplicateFileId(entry.file_id,
1099
self._byid[entry.file_id])
1101
if entry.parent_id is None:
1102
assert self.root is None and len(self._byid) == 0
1103
self._set_root(entry)
396
1106
parent = self._byid[entry.parent_id]
397
1107
except KeyError:
398
bailout("parent_id {%s} not in inventory" % entry.parent_id)
1108
raise BzrError("parent_id {%s} not in inventory" % entry.parent_id)
400
if parent.children.has_key(entry.name):
401
bailout("%s is already versioned" %
402
appendpath(self.id2path(parent.file_id), entry.name))
1110
if entry.name in parent.children:
1111
raise BzrError("%s is already versioned" %
1112
osutils.pathjoin(self.id2path(parent.file_id), entry.name))
404
1114
self._byid[entry.file_id] = entry
405
1115
parent.children[entry.name] = entry
408
def add_path(self, relpath, kind, file_id=None):
1118
def add_path(self, relpath, kind, file_id=None, parent_id=None):
409
1119
"""Add entry from a path.
411
The immediate parent must already be versioned"""
412
parts = bzrlib.osutils.splitpath(relpath)
1121
The immediate parent must already be versioned.
1123
Returns the new entry object."""
1125
parts = osutils.splitpath(relpath)
413
1127
if len(parts) == 0:
414
bailout("cannot re-add root of inventory")
417
file_id = bzrlib.branch.gen_file_id(relpath)
419
parent_id = self.path2id(parts[:-1])
420
assert parent_id != None
421
ie = InventoryEntry(file_id, parts[-1],
422
kind=kind, parent_id=parent_id)
1129
file_id = generate_ids.gen_root_id()
1131
file_id = osutils.safe_file_id(file_id)
1132
self.root = InventoryDirectory(file_id, '', None)
1133
self._byid = {self.root.file_id: self.root}
1136
parent_path = parts[:-1]
1137
parent_id = self.path2id(parent_path)
1138
if parent_id is None:
1139
raise errors.NotVersionedError(path=parent_path)
1140
ie = make_entry(kind, parts[-1], parent_id, file_id)
423
1141
return self.add(ie)
426
1143
def __delitem__(self, file_id):
427
1144
"""Remove entry by id.
429
1146
>>> inv = Inventory()
430
>>> inv.add(InventoryEntry('123', 'foo.c', 'file', ROOT_ID))
1147
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1148
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
431
1149
>>> '123' in inv
433
1151
>>> del inv['123']
434
1152
>>> '123' in inv
1155
file_id = osutils.safe_file_id(file_id)
437
1156
ie = self[file_id]
439
assert self[ie.parent_id].children[ie.name] == ie
1158
assert ie.parent_id is None or \
1159
self[ie.parent_id].children[ie.name] == ie
441
# TODO: Test deleting all children; maybe hoist to a separate
443
if ie.kind == 'directory':
444
for cie in ie.children.values():
445
del self[cie.file_id]
448
1161
del self._byid[file_id]
449
del self[ie.parent_id].children[ie.name]
453
return Set(self._byid)
456
def to_element(self):
457
"""Convert to XML Element"""
458
e = Element('inventory')
460
for path, ie in self.iter_entries():
461
e.append(ie.to_element())
465
def from_element(cls, elt):
466
"""Construct from XML Element
468
>>> inv = Inventory()
469
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c', 'file', ROOT_ID))
470
>>> elt = inv.to_element()
471
>>> inv2 = Inventory.from_element(elt)
475
assert elt.tag == 'inventory'
478
o.add(InventoryEntry.from_element(e))
481
from_element = classmethod(from_element)
484
def __cmp__(self, other):
1162
if ie.parent_id is not None:
1163
del self[ie.parent_id].children[ie.name]
1165
def __eq__(self, other):
485
1166
"""Compare two sets by comparing their contents.
487
1168
>>> i1 = Inventory()
488
1169
>>> i2 = Inventory()
491
>>> i1.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
1172
>>> i1.add(InventoryFile('123', 'foo', ROOT_ID))
1173
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
494
>>> i2.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
1176
>>> i2.add(InventoryFile('123', 'foo', ROOT_ID))
1177
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
501
1181
if not isinstance(other, Inventory):
502
1182
return NotImplemented
504
if self.id_set() ^ other.id_set():
507
for file_id in self._byid:
508
c = cmp(self[file_id], other[file_id])
1184
return self._byid == other._byid
1186
def __ne__(self, other):
1187
return not self.__eq__(other)
1190
raise ValueError('not hashable')
1192
def _iter_file_id_parents(self, file_id):
1193
"""Yield the parents of file_id up to the root."""
1194
file_id = osutils.safe_file_id(file_id)
1195
while file_id is not None:
1197
ie = self._byid[file_id]
1199
raise BzrError("file_id {%s} not found in inventory" % file_id)
1201
file_id = ie.parent_id
514
1203
def get_idpath(self, file_id):
515
1204
"""Return a list of file_ids for the path to an entry.
added
removed
bzrlib/inventory.py
