« back to all changes in this revision
Viewing changes to bzrlib/repository.py
- browse files at revision 1508.1.19
- compare with another revision
- download diff
- download tarball
- view history from revision 1508.1.19
- Committer: Robert Collins
- Date: 2006-02-11 11:58:06 UTC
- mto: (1534.1.22 integration)
- mto: This revision was merged to the branch mainline in revision 1554.
- Revision ID: robertc@robertcollins.net-20060211115806-732dabc1e35714ed
Give format3 working trees their own last-revision marker.
- files added:
- bzrlib/_merge_core.py
- bzrlib/util/urlgrabber
- notes
- patches
- files removed:
- COPYING.txt
- bzrlib/benchmarks
- bzrlib/benchmarks/tree_creator
- bzrlib/bundle
- bzrlib/bundle/old
- bzrlib/bundle/serializer
- bzrlib/plugins/launchpad
- bzrlib/store/revision
- bzrlib/store/versioned
- bzrlib/tests/interrepository_implementations
- bzrlib/tests/intertree_implementations
- bzrlib/tests/interversionedfile_implementations
- bzrlib/tests/revisionstore_implementations
- bzrlib/tests/test_patches_data
- bzrlib/tests/tree_implementations
- bzrlib/transport/http
- tools/win32
- files renamed:
- bzrlib/bundle/__init__.py => bzrlib/_changeset.py
- bzrlib/store/versioned/__init__.py => bzrlib/store/weave.py
- bzrlib/tests/blackbox/test_annotate.py => bzrlib/tests/test_annotate.py
- bzrlib/tests/workingtree_implementations/test_basis_inventory.py => bzrlib/tests/test_basis_inventory.py
- bzrlib/tests/repository_implementations/test_revision.py => bzrlib/tests/test_revprops.py
- bzrlib/tests/blackbox/test_uncommit.py => bzrlib/tests/test_uncommit.py
- bzrlib/transport/http/__init__.py => bzrlib/transport/http.py
- doc/tutorial.txt => tutorial.txt
- files modified:
- .bzrignore
added
removed
bzrlib/repository.py
1
# Copyright (C) 2005, 2006 Canonical Ltd
2
#
1
# Copyright (C) 2005 Canonical Ltd
2
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.
7
#
7
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.
12
#
12
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
16
16
17
from binascii import hexlify
18
17
from copy import deepcopy
19
18
from cStringIO import StringIO
20
import re
21
import time
22
19
from unittest import TestSuite
23
24
from bzrlib import bzrdir, check, delta, gpg, errors, xml5, ui, transactions, osutils
20
import xml.sax.saxutils
21
22
23
import bzrlib.bzrdir as bzrdir
25
24
from bzrlib.decorators import needs_read_lock, needs_write_lock
25
import bzrlib.errors as errors
26
26
from bzrlib.errors import InvalidRevisionId
27
from bzrlib.graph import Graph
28
from bzrlib.inter import InterObject
29
from bzrlib.inventory import Inventory, InventoryDirectory, ROOT_ID
30
from bzrlib.knit import KnitVersionedFile, KnitPlainFactory
31
from bzrlib.lockable_files import LockableFiles, TransportLock
32
from bzrlib.lockdir import LockDir
33
from bzrlib.osutils import (safe_unicode, rand_bytes, compact_date,
34
local_time_offset)
35
from bzrlib.revision import NULL_REVISION, Revision
36
from bzrlib.revisiontree import RevisionTree
37
from bzrlib.store.versioned import VersionedFileStore, WeaveStore
27
from bzrlib.lockable_files import LockableFiles
28
from bzrlib.osutils import safe_unicode
29
from bzrlib.revision import NULL_REVISION
30
from bzrlib.store import copy_all
31
from bzrlib.store.weave import WeaveStore
38
32
from bzrlib.store.text import TextStore
39
from bzrlib import symbol_versioning
40
from bzrlib.symbol_versioning import (deprecated_method,
41
zero_nine,
42
)
33
from bzrlib.symbol_versioning import *
34
from bzrlib.trace import mutter
35
from bzrlib.tree import RevisionTree
43
36
from bzrlib.testament import Testament
44
from bzrlib.trace import mutter, note, warning
45
from bzrlib.tsort import topo_sort
46
from bzrlib.weave import WeaveFile
47
48
49
# Old formats display a warning, but only once
50
_deprecation_warning_done = False
37
from bzrlib.tree import EmptyTree
38
import bzrlib.xml5
51
39
52
40
53
41
class Repository(object):
62
50
remote) disk.
63
51
"""
64
52
65
@needs_write_lock
66
def add_inventory(self, revid, inv, parents):
67
"""Add the inventory inv to the repository as revid.
68
69
:param parents: The revision ids of the parents that revid
70
is known to have and are in the repository already.
71
72
returns the sha1 of the serialized inventory.
73
"""
74
assert inv.revision_id is None or inv.revision_id == revid, \
75
"Mismatch between inventory revision" \
76
" id and insertion revid (%r, %r)" % (inv.revision_id, revid)
77
assert inv.root is not None
78
inv_text = xml5.serializer_v5.write_inventory_to_string(inv)
79
inv_sha1 = osutils.sha_string(inv_text)
80
inv_vf = self.control_weaves.get_weave('inventory',
81
self.get_transaction())
82
self._inventory_add_lines(inv_vf, revid, parents, osutils.split_lines(inv_text))
83
return inv_sha1
84
85
def _inventory_add_lines(self, inv_vf, revid, parents, lines):
86
final_parents = []
87
for parent in parents:
88
if parent in inv_vf:
89
final_parents.append(parent)
90
91
inv_vf.add_lines(revid, final_parents, lines)
92
93
@needs_write_lock
94
def add_revision(self, rev_id, rev, inv=None, config=None):
95
"""Add rev to the revision store as rev_id.
96
97
:param rev_id: the revision id to use.
98
:param rev: The revision object.
99
:param inv: The inventory for the revision. if None, it will be looked
100
up in the inventory storer
101
:param config: If None no digital signature will be created.
102
If supplied its signature_needed method will be used
103
to determine if a signature should be made.
104
"""
105
if config is not None and config.signature_needed():
106
if inv is None:
107
inv = self.get_inventory(rev_id)
108
plaintext = Testament(rev, inv).as_short_text()
109
self.store_revision_signature(
110
gpg.GPGStrategy(config), plaintext, rev_id)
111
if not rev_id in self.get_inventory_weave():
112
if inv is None:
113
raise errors.WeaveRevisionNotPresent(rev_id,
114
self.get_inventory_weave())
115
else:
116
# yes, this is not suitable for adding with ghosts.
117
self.add_inventory(rev_id, inv, rev.parent_ids)
118
self._revision_store.add_revision(rev, self.get_transaction())
119
120
53
@needs_read_lock
121
54
def _all_possible_ids(self):
122
55
"""Return all the possible revisions that we could find."""
123
return self.get_inventory_weave().versions()
56
return self.get_inventory_weave().names()
124
57
58
@needs_read_lock
125
59
def all_revision_ids(self):
126
60
"""Returns a list of all the revision ids in the repository.
127
61
128
This is deprecated because code should generally work on the graph
129
reachable from a particular revision, and ignore any other revisions
130
that might be present. There is no direct replacement method.
131
"""
132
return self._all_revision_ids()
133
134
@needs_read_lock
135
def _all_revision_ids(self):
136
"""Returns a list of all the revision ids in the repository.
137
138
62
These are in as much topological order as the underlying store can
139
63
present: for weaves ghosts may lead to a lack of correctness until
140
64
the reweave updates the parents list.
141
65
"""
142
if self._revision_store.text_store.listable():
143
return self._revision_store.all_revision_ids(self.get_transaction())
144
66
result = self._all_possible_ids()
145
67
return self._eliminate_revisions_not_present(result)
146
68
147
def break_lock(self):
148
"""Break a lock if one is present from another instance.
149
150
Uses the ui factory to ask for confirmation if the lock may be from
151
an active process.
152
"""
153
self.control_files.break_lock()
154
155
69
@needs_read_lock
156
70
def _eliminate_revisions_not_present(self, revision_ids):
157
71
"""Check every revision id in revision_ids to see if we have it.
169
83
"""Construct the current default format repository in a_bzrdir."""
170
84
return RepositoryFormat.get_default_format().initialize(a_bzrdir)
171
85
172
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
173
"""instantiate a Repository.
174
175
:param _format: The format of the repository on disk.
176
:param a_bzrdir: The BzrDir of the repository.
177
178
In the future we will have a single api for all stores for
179
getting file texts, inventories and revisions, then
180
this construct will accept instances of those things.
181
"""
182
super(Repository, self).__init__()
86
def __init__(self, transport, branch_format, _format=None, a_bzrdir=None):
87
object.__init__(self)
88
if transport is not None:
89
warn("Repository.__init__(..., transport=XXX): The transport parameter is "
90
"deprecated and was never in a supported release. Please use "
91
"bzrdir.open_repository() or bzrdir.open_branch().repository.",
92
DeprecationWarning,
93
stacklevel=2)
94
self.control_files = LockableFiles(transport.clone(bzrlib.BZRDIR), 'README')
95
else:
96
# TODO: clone into repository if needed
97
self.control_files = LockableFiles(a_bzrdir.get_repository_transport(None), 'README')
98
99
dir_mode = self.control_files._dir_mode
100
file_mode = self.control_files._file_mode
183
101
self._format = _format
184
# the following are part of the public API for Repository:
185
102
self.bzrdir = a_bzrdir
186
self.control_files = control_files
187
self._revision_store = _revision_store
188
self.text_store = text_store
189
# backwards compatibility
190
self.weave_store = text_store
191
# not right yet - should be more semantically clear ?
192
#
193
self.control_store = control_store
194
self.control_weaves = control_store
195
# TODO: make sure to construct the right store classes, etc, depending
196
# on whether escaping is required.
197
self._warn_if_deprecated()
198
199
def __repr__(self):
200
return '%s(%r)' % (self.__class__.__name__,
201
self.bzrdir.transport.base)
202
203
def is_locked(self):
204
return self.control_files.is_locked()
103
104
def get_weave(name, prefixed=False):
105
if name:
106
name = safe_unicode(name)
107
else:
108
name = ''
109
relpath = self.control_files._escape(name)
110
weave_transport = self.control_files._transport.clone(relpath)
111
ws = WeaveStore(weave_transport, prefixed=prefixed,
112
dir_mode=dir_mode,
113
file_mode=file_mode)
114
if self.control_files._transport.should_cache():
115
ws.enable_cache = True
116
return ws
117
118
119
def get_store(name, compressed=True, prefixed=False):
120
# FIXME: This approach of assuming stores are all entirely compressed
121
# or entirely uncompressed is tidy, but breaks upgrade from
122
# some existing branches where there's a mixture; we probably
123
# still want the option to look for both.
124
if name:
125
name = safe_unicode(name)
126
else:
127
name = ''
128
relpath = self.control_files._escape(name)
129
store = TextStore(self.control_files._transport.clone(relpath),
130
prefixed=prefixed, compressed=compressed,
131
dir_mode=dir_mode,
132
file_mode=file_mode)
133
#if self._transport.should_cache():
134
# cache_path = os.path.join(self.cache_root, name)
135
# os.mkdir(cache_path)
136
# store = bzrlib.store.CachedStore(store, cache_path)
137
return store
138
139
if branch_format is not None:
140
# circular dependencies:
141
from bzrlib.branch import (BzrBranchFormat4,
142
BzrBranchFormat5,
143
BzrBranchFormat6,
144
)
145
if isinstance(branch_format, BzrBranchFormat4):
146
self._format = RepositoryFormat4()
147
elif isinstance(branch_format, BzrBranchFormat5):
148
self._format = RepositoryFormat5()
149
elif isinstance(branch_format, BzrBranchFormat6):
150
self._format = RepositoryFormat6()
151
152
153
if isinstance(self._format, RepositoryFormat4):
154
self.inventory_store = get_store('inventory-store')
155
self.text_store = get_store('text-store')
156
self.revision_store = get_store('revision-store')
157
elif isinstance(self._format, RepositoryFormat5):
158
self.control_weaves = get_weave('')
159
self.weave_store = get_weave('weaves')
160
self.revision_store = get_store('revision-store', compressed=False)
161
elif isinstance(self._format, RepositoryFormat6):
162
self.control_weaves = get_weave('')
163
self.weave_store = get_weave('weaves', prefixed=True)
164
self.revision_store = get_store('revision-store', compressed=False,
165
prefixed=True)
166
elif isinstance(self._format, RepositoryFormat7):
167
self.control_weaves = get_weave('')
168
self.weave_store = get_weave('weaves', prefixed=True)
169
self.revision_store = get_store('revision-store', compressed=False,
170
prefixed=True)
171
self.revision_store.register_suffix('sig')
205
172
206
173
def lock_write(self):
207
174
self.control_files.lock_write()
209
176
def lock_read(self):
210
177
self.control_files.lock_read()
211
178
212
def get_physical_lock_status(self):
213
return self.control_files.get_physical_lock_status()
214
215
179
@needs_read_lock
216
180
def missing_revision_ids(self, other, revision_id=None):
217
181
"""Return the revision ids that other has that this does not.
220
184
221
185
revision_id: only return revision ids included by revision_id.
222
186
"""
223
return InterRepository.get(other, self).missing_revision_ids(revision_id)
187
if self._compatible_formats(other):
188
# fast path for weave-inventory based stores.
189
# we want all revisions to satisft revision_id in other.
190
# but we dont want to stat every file here and there.
191
# we want then, all revisions other needs to satisfy revision_id
192
# checked, but not those that we have locally.
193
# so the first thing is to get a subset of the revisions to
194
# satisfy revision_id in other, and then eliminate those that
195
# we do already have.
196
# this is slow on high latency connection to self, but as as this
197
# disk format scales terribly for push anyway due to rewriting
198
# inventory.weave, this is considered acceptable.
199
# - RBC 20060209
200
if revision_id is not None:
201
other_ids = other.get_ancestry(revision_id)
202
assert other_ids.pop(0) == None
203
else:
204
other_ids = other._all_possible_ids()
205
other_ids_set = set(other_ids)
206
# other ids is the worst case to pull now.
207
# now we want to filter other_ids against what we actually
208
# have, but dont try to stat what we know we dont.
209
my_ids = set(self._all_possible_ids())
210
possibly_present_revisions = my_ids.intersection(other_ids_set)
211
actually_present_revisions = set(self._eliminate_revisions_not_present(possibly_present_revisions))
212
required_revisions = other_ids_set.difference(actually_present_revisions)
213
required_topo_revisions = [rev_id for rev_id in other_ids if rev_id in required_revisions]
214
if revision_id is not None:
215
# we used get_ancestry to determine other_ids then we are assured all
216
# revisions referenced are present as they are installed in topological order.
217
return required_topo_revisions
218
else:
219
# we only have an estimate of whats available
220
return other._eliminate_revisions_not_present(required_topo_revisions)
221
# slow code path.
222
my_ids = set(self.all_revision_ids())
223
if revision_id is not None:
224
other_ids = other.get_ancestry(revision_id)
225
assert other_ids.pop(0) == None
226
else:
227
other_ids = other.all_revision_ids()
228
result_set = set(other_ids).difference(my_ids)
229
return [rev_id for rev_id in other_ids if rev_id in result_set]
224
230
225
231
@staticmethod
226
232
def open(base):
232
238
control = bzrdir.BzrDir.open(base)
233
239
return control.open_repository()
234
240
241
def _compatible_formats(self, other):
242
"""Return True if the stores in self and other are 'compatible'
243
244
'compatible' means that they are both the same underlying type
245
i.e. both weave stores, or both knits and thus support fast-path
246
operations."""
247
return (isinstance(self._format, (RepositoryFormat5,
248
RepositoryFormat6,
249
RepositoryFormat7)) and
250
isinstance(other._format, (RepositoryFormat5,
251
RepositoryFormat6,
252
RepositoryFormat7)))
253
254
@needs_read_lock
235
255
def copy_content_into(self, destination, revision_id=None, basis=None):
236
"""Make a complete copy of the content in self into destination.
237
238
This is a destructive operation! Do not use it on existing
239
repositories.
240
"""
241
return InterRepository.get(self, destination).copy_content(revision_id, basis)
256
"""Make a complete copy of the content in self into destination."""
257
destination.lock_write()
258
try:
259
# optimised paths:
260
# compatible stores
261
if self._compatible_formats(destination):
262
if basis is not None:
263
# copy the basis in, then fetch remaining data.
264
basis.copy_content_into(destination, revision_id)
265
destination.fetch(self, revision_id=revision_id)
266
else:
267
# FIXME do not peek!
268
if self.control_files._transport.listable():
269
destination.control_weaves.copy_multi(self.control_weaves,
270
['inventory'])
271
copy_all(self.weave_store, destination.weave_store)
272
copy_all(self.revision_store, destination.revision_store)
273
else:
274
destination.fetch(self, revision_id=revision_id)
275
# compatible v4 stores
276
elif isinstance(self._format, RepositoryFormat4):
277
if not isinstance(destination._format, RepositoryFormat4):
278
raise BzrError('cannot copy v4 branches to anything other than v4 branches.')
279
store_pairs = ((self.text_store, destination.text_store),
280
(self.inventory_store, destination.inventory_store),
281
(self.revision_store, destination.revision_store))
282
try:
283
for from_store, to_store in store_pairs:
284
copy_all(from_store, to_store)
285
except UnlistableStore:
286
raise UnlistableBranch(from_store)
287
# fallback - 'fetch'
288
else:
289
destination.fetch(self, revision_id=revision_id)
290
finally:
291
destination.unlock()
242
292
243
def fetch(self, source, revision_id=None, pb=None):
293
@needs_write_lock
294
def fetch(self, source, revision_id=None):
244
295
"""Fetch the content required to construct revision_id from source.
245
296
246
297
If revision_id is None all content is copied.
247
298
"""
248
return InterRepository.get(source, self).fetch(revision_id=revision_id,
249
pb=pb)
250
251
def get_commit_builder(self, branch, parents, config, timestamp=None,
252
timezone=None, committer=None, revprops=None,
253
revision_id=None):
254
"""Obtain a CommitBuilder for this repository.
255
256
:param branch: Branch to commit to.
257
:param parents: Revision ids of the parents of the new revision.
258
:param config: Configuration to use.
259
:param timestamp: Optional timestamp recorded for commit.
260
:param timezone: Optional timezone for timestamp.
261
:param committer: Optional committer to set for commit.
262
:param revprops: Optional dictionary of revision properties.
263
:param revision_id: Optional revision id.
264
"""
265
return _CommitBuilder(self, parents, config, timestamp, timezone,
266
committer, revprops, revision_id)
299
from bzrlib.fetch import RepoFetcher
300
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
301
source, source._format, self, self._format)
302
RepoFetcher(to_repository=self, from_repository=source, last_revision=revision_id)
267
303
268
304
def unlock(self):
269
305
self.control_files.unlock()
285
321
bzrdir.BzrDirFormat6)):
286
322
result = a_bzrdir.open_repository()
287
323
else:
288
result = self._format.initialize(a_bzrdir, shared=self.is_shared())
324
result = self._format.initialize(a_bzrdir)
289
325
self.copy_content_into(result, revision_id, basis)
290
326
return result
291
327
292
@needs_read_lock
293
328
def has_revision(self, revision_id):
294
"""True if this repository has a copy of the revision."""
295
return self._revision_store.has_revision_id(revision_id,
296
self.get_transaction())
329
"""True if this branch has a copy of the revision.
330
331
This does not necessarily imply the revision is merge
332
or on the mainline."""
333
return (revision_id is None
334
or self.revision_store.has_id(revision_id))
297
335
298
336
@needs_read_lock
299
def get_revision_reconcile(self, revision_id):
300
"""'reconcile' helper routine that allows access to a revision always.
301
302
This variant of get_revision does not cross check the weave graph
303
against the revision one as get_revision does: but it should only
304
be used by reconcile, or reconcile-alike commands that are correcting
305
or testing the revision graph.
306
"""
337
def get_revision_xml_file(self, revision_id):
338
"""Return XML file object for revision object."""
307
339
if not revision_id or not isinstance(revision_id, basestring):
308
340
raise InvalidRevisionId(revision_id=revision_id, branch=self)
309
return self._revision_store.get_revisions([revision_id],
310
self.get_transaction())[0]
311
@needs_read_lock
312
def get_revisions(self, revision_ids):
313
return self._revision_store.get_revisions(revision_ids,
314
self.get_transaction())
341
try:
342
return self.revision_store.get(revision_id)
343
except (IndexError, KeyError):
344
raise bzrlib.errors.NoSuchRevision(self, revision_id)
315
345
316
346
@needs_read_lock
317
347
def get_revision_xml(self, revision_id):
318
rev = self.get_revision(revision_id)
319
rev_tmp = StringIO()
320
# the current serializer..
321
self._revision_store._serializer.write_revision(rev, rev_tmp)
322
rev_tmp.seek(0)
323
return rev_tmp.getvalue()
348
return self.get_revision_xml_file(revision_id).read()
324
349
325
350
@needs_read_lock
326
351
def get_revision(self, revision_id):
327
352
"""Return the Revision object for a named revision"""
328
r = self.get_revision_reconcile(revision_id)
329
# weave corruption can lead to absent revision markers that should be
330
# present.
331
# the following test is reasonably cheap (it needs a single weave read)
332
# and the weave is cached in read transactions. In write transactions
333
# it is not cached but typically we only read a small number of
334
# revisions. For knits when they are introduced we will probably want
335
# to ensure that caching write transactions are in use.
336
inv = self.get_inventory_weave()
337
self._check_revision_parents(r, inv)
353
xml_file = self.get_revision_xml_file(revision_id)
354
355
try:
356
r = bzrlib.xml5.serializer_v5.read_revision(xml_file)
357
except SyntaxError, e:
358
raise bzrlib.errors.BzrError('failed to unpack revision_xml',
359
[revision_id,
360
str(e)])
361
362
assert r.revision_id == revision_id
338
363
return r
339
364
340
365
@needs_read_lock
341
def get_deltas_for_revisions(self, revisions):
342
"""Produce a generator of revision deltas.
343
344
Note that the input is a sequence of REVISIONS, not revision_ids.
345
Trees will be held in memory until the generator exits.
346
Each delta is relative to the revision's lefthand predecessor.
347
"""
348
required_trees = set()
349
for revision in revisions:
350
required_trees.add(revision.revision_id)
351
required_trees.update(revision.parent_ids[:1])
352
trees = dict((t.get_revision_id(), t) for
353
t in self.revision_trees(required_trees))
354
for revision in revisions:
355
if not revision.parent_ids:
356
old_tree = self.revision_tree(None)
357
else:
358
old_tree = trees[revision.parent_ids[0]]
359
yield trees[revision.revision_id].changes_from(old_tree)
360
361
@needs_read_lock
362
def get_revision_delta(self, revision_id):
363
"""Return the delta for one revision.
364
365
The delta is relative to the left-hand predecessor of the
366
revision.
367
"""
368
r = self.get_revision(revision_id)
369
return list(self.get_deltas_for_revisions([r]))[0]
370
371
def _check_revision_parents(self, revision, inventory):
372
"""Private to Repository and Fetch.
373
374
This checks the parentage of revision in an inventory weave for
375
consistency and is only applicable to inventory-weave-for-ancestry
376
using repository formats & fetchers.
377
"""
378
weave_parents = inventory.get_parents(revision.revision_id)
379
weave_names = inventory.versions()
380
for parent_id in revision.parent_ids:
381
if parent_id in weave_names:
382
# this parent must not be a ghost.
383
if not parent_id in weave_parents:
384
# but it is a ghost
385
raise errors.CorruptRepository(self)
366
def get_revision_sha1(self, revision_id):
367
"""Hash the stored value of a revision, and return it."""
368
# In the future, revision entries will be signed. At that
369
# point, it is probably best *not* to include the signature
370
# in the revision hash. Because that lets you re-sign
371
# the revision, (add signatures/remove signatures) and still
372
# have all hash pointers stay consistent.
373
# But for now, just hash the contents.
374
return bzrlib.osutils.sha_file(self.get_revision_xml_file(revision_id))
386
375
387
376
@needs_write_lock
388
377
def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
389
signature = gpg_strategy.sign(plaintext)
390
self._revision_store.add_revision_signature_text(revision_id,
391
signature,
392
self.get_transaction())
393
394
def fileids_altered_by_revision_ids(self, revision_ids):
395
"""Find the file ids and versions affected by revisions.
396
397
:param revisions: an iterable containing revision ids.
398
:return: a dictionary mapping altered file-ids to an iterable of
399
revision_ids. Each altered file-ids has the exact revision_ids that
400
altered it listed explicitly.
378
self.revision_store.add(StringIO(gpg_strategy.sign(plaintext)),
379
revision_id, "sig")
380
381
def fileid_involved_between_revs(self, from_revid, to_revid):
382
"""Find file_id(s) which are involved in the changes between revisions.
383
384
This determines the set of revisions which are involved, and then
385
finds all file ids affected by those revisions.
386
"""
387
# TODO: jam 20060119 This code assumes that w.inclusions will
388
# always be correct. But because of the presence of ghosts
389
# it is possible to be wrong.
390
# One specific example from Robert Collins:
391
# Two branches, with revisions ABC, and AD
392
# C is a ghost merge of D.
393
# Inclusions doesn't recognize D as an ancestor.
394
# If D is ever merged in the future, the weave
395
# won't be fixed, because AD never saw revision C
396
# to cause a conflict which would force a reweave.
397
w = self.get_inventory_weave()
398
from_set = set(w.inclusions([w.lookup(from_revid)]))
399
to_set = set(w.inclusions([w.lookup(to_revid)]))
400
included = to_set.difference(from_set)
401
changed = map(w.idx_to_name, included)
402
return self._fileid_involved_by_set(changed)
403
404
def fileid_involved(self, last_revid=None):
405
"""Find all file_ids modified in the ancestry of last_revid.
406
407
:param last_revid: If None, last_revision() will be used.
408
"""
409
w = self.get_inventory_weave()
410
if not last_revid:
411
changed = set(w._names)
412
else:
413
included = w.inclusions([w.lookup(last_revid)])
414
changed = map(w.idx_to_name, included)
415
return self._fileid_involved_by_set(changed)
416
417
def fileid_involved_by_set(self, changes):
418
"""Find all file_ids modified by the set of revisions passed in.
419
420
:param changes: A set() of revision ids
421
"""
422
# TODO: jam 20060119 This line does *nothing*, remove it.
423
# or better yet, change _fileid_involved_by_set so
424
# that it takes the inventory weave, rather than
425
# pulling it out by itself.
426
return self._fileid_involved_by_set(changes)
427
428
def _fileid_involved_by_set(self, changes):
429
"""Find the set of file-ids affected by the set of revisions.
430
431
:param changes: A set() of revision ids.
432
:return: A set() of file ids.
433
434
This peaks at the Weave, interpreting each line, looking to
435
see if it mentions one of the revisions. And if so, includes
436
the file id mentioned.
437
This expects both the Weave format, and the serialization
438
to have a single line per file/directory, and to have
439
fileid="" and revision="" on that line.
401
440
"""
402
441
assert isinstance(self._format, (RepositoryFormat5,
403
442
RepositoryFormat6,
404
RepositoryFormat7,
405
RepositoryFormatKnit1)), \
406
("fileids_altered_by_revision_ids only supported for branches "
407
"which store inventory as unnested xml, not on %r" % self)
408
selected_revision_ids = set(revision_ids)
443
RepositoryFormat7)), \
444
"fileid_involved only supported for branches which store inventory as unnested xml"
445
409
446
w = self.get_inventory_weave()
410
result = {}
411
412
# this code needs to read every new line in every inventory for the
413
# inventories [revision_ids]. Seeing a line twice is ok. Seeing a line
414
# not present in one of those inventories is unnecessary but not
415
# harmful because we are filtering by the revision id marker in the
416
# inventory lines : we only select file ids altered in one of those
417
# revisions. We don't need to see all lines in the inventory because
418
# only those added in an inventory in rev X can contain a revision=X
419
# line.
420
for line in w.iter_lines_added_or_present_in_versions(selected_revision_ids):
447
file_ids = set()
448
for line in w._weave:
449
450
# it is ugly, but it is due to the weave structure
451
if not isinstance(line, basestring): continue
452
421
453
start = line.find('file_id="')+9
422
454
if start < 9: continue
423
455
end = line.find('"', start)
424
456
assert end>= 0
425
file_id = _unescape_xml(line[start:end])
457
file_id = xml.sax.saxutils.unescape(line[start:end])
458
459
# check if file_id is already present
460
if file_id in file_ids: continue
426
461
427
462
start = line.find('revision="')+10
428
463
if start < 10: continue
429
464
end = line.find('"', start)
430
465
assert end>= 0
431
revision_id = _unescape_xml(line[start:end])
432
if revision_id in selected_revision_ids:
433
result.setdefault(file_id, set()).add(revision_id)
434
return result
466
revision_id = xml.sax.saxutils.unescape(line[start:end])
467
468
if revision_id in changes:
469
file_ids.add(file_id)
470
return file_ids
435
471
436
472
@needs_read_lock
437
473
def get_inventory_weave(self):
441
477
@needs_read_lock
442
478
def get_inventory(self, revision_id):
443
479
"""Get Inventory object by hash."""
444
return self.deserialise_inventory(
445
revision_id, self.get_inventory_xml(revision_id))
446
447
def deserialise_inventory(self, revision_id, xml):
448
"""Transform the xml into an inventory object.
449
450
:param revision_id: The expected revision id of the inventory.
451
:param xml: A serialised inventory.
452
"""
453
result = xml5.serializer_v5.read_inventory_from_string(xml)
454
result.root.revision = revision_id
455
return result
480
xml = self.get_inventory_xml(revision_id)
481
return bzrlib.xml5.serializer_v5.read_inventory_from_string(xml)
456
482
457
483
@needs_read_lock
458
484
def get_inventory_xml(self, revision_id):
460
486
try:
461
487
assert isinstance(revision_id, basestring), type(revision_id)
462
488
iw = self.get_inventory_weave()
463
return iw.get_text(revision_id)
489
return iw.get_text(iw.lookup(revision_id))
464
490
except IndexError:
465
raise errors.HistoryMissing(self, 'inventory', revision_id)
491
raise bzrlib.errors.HistoryMissing(self, 'inventory', revision_id)
466
492
467
493
@needs_read_lock
468
494
def get_inventory_sha1(self, revision_id):
471
497
return self.get_revision(revision_id).inventory_sha1
472
498
473
499
@needs_read_lock
474
def get_revision_graph(self, revision_id=None):
475
"""Return a dictionary containing the revision graph.
476
477
:param revision_id: The revision_id to get a graph from. If None, then
478
the entire revision graph is returned. This is a deprecated mode of
479
operation and will be removed in the future.
480
:return: a dictionary of revision_id->revision_parents_list.
481
"""
482
# special case NULL_REVISION
483
if revision_id == NULL_REVISION:
484
return {}
485
weave = self.get_inventory_weave()
486
all_revisions = self._eliminate_revisions_not_present(weave.versions())
487
entire_graph = dict([(node, weave.get_parents(node)) for
488
node in all_revisions])
489
if revision_id is None:
490
return entire_graph
491
elif revision_id not in entire_graph:
492
raise errors.NoSuchRevision(self, revision_id)
493
else:
494
# add what can be reached from revision_id
495
result = {}
496
pending = set([revision_id])
497
while len(pending) > 0:
498
node = pending.pop()
499
result[node] = entire_graph[node]
500
for revision_id in result[node]:
501
if revision_id not in result:
502
pending.add(revision_id)
503
return result
504
505
@needs_read_lock
506
def get_revision_graph_with_ghosts(self, revision_ids=None):
507
"""Return a graph of the revisions with ghosts marked as applicable.
508
509
:param revision_ids: an iterable of revisions to graph or None for all.
510
:return: a Graph object with the graph reachable from revision_ids.
511
"""
512
result = Graph()
513
if not revision_ids:
514
pending = set(self.all_revision_ids())
515
required = set([])
516
else:
517
pending = set(revision_ids)
518
# special case NULL_REVISION
519
if NULL_REVISION in pending:
520
pending.remove(NULL_REVISION)
521
required = set(pending)
522
done = set([])
523
while len(pending):
524
revision_id = pending.pop()
525
try:
526
rev = self.get_revision(revision_id)
527
except errors.NoSuchRevision:
528
if revision_id in required:
529
raise
530
# a ghost
531
result.add_ghost(revision_id)
532
continue
533
for parent_id in rev.parent_ids:
534
# is this queued or done ?
535
if (parent_id not in pending and
536
parent_id not in done):
537
# no, queue it.
538
pending.add(parent_id)
539
result.add_node(revision_id, rev.parent_ids)
540
done.add(revision_id)
541
return result
542
543
@needs_read_lock
544
500
def get_revision_inventory(self, revision_id):
545
501
"""Return inventory of a past revision."""
546
502
# TODO: Unify this with get_inventory()
557
513
return self.get_inventory(revision_id)
558
514
559
515
@needs_read_lock
560
def is_shared(self):
561
"""Return True if this repository is flagged as a shared repository."""
562
raise NotImplementedError(self.is_shared)
563
564
@needs_write_lock
565
def reconcile(self, other=None, thorough=False):
566
"""Reconcile this repository."""
567
from bzrlib.reconcile import RepoReconciler
568
reconciler = RepoReconciler(self, thorough=thorough)
569
reconciler.reconcile()
570
return reconciler
571
572
@needs_read_lock
573
516
def revision_tree(self, revision_id):
574
517
"""Return Tree for a revision on this branch.
575
518
576
`revision_id` may be None for the empty tree revision.
577
"""
519
`revision_id` may be None for the null revision, in which case
520
an `EmptyTree` is returned."""
578
521
# TODO: refactor this to use an existing revision object
579
522
# so we don't need to read it in twice.
580
523
if revision_id is None or revision_id == NULL_REVISION:
581
return RevisionTree(self, Inventory(), NULL_REVISION)
524
return EmptyTree()
582
525
else:
583
526
inv = self.get_revision_inventory(revision_id)
584
527
return RevisionTree(self, inv, revision_id)
585
528
586
529
@needs_read_lock
587
def revision_trees(self, revision_ids):
588
"""Return Tree for a revision on this branch.
589
590
`revision_id` may not be None or 'null:'"""
591
assert None not in revision_ids
592
assert NULL_REVISION not in revision_ids
593
texts = self.get_inventory_weave().get_texts(revision_ids)
594
for text, revision_id in zip(texts, revision_ids):
595
inv = self.deserialise_inventory(revision_id, text)
596
yield RevisionTree(self, inv, revision_id)
597
598
@needs_read_lock
599
530
def get_ancestry(self, revision_id):
600
531
"""Return a list of revision-ids integrated by a revision.
601
602
The first element of the list is always None, indicating the origin
603
revision. This might change when we have history horizons, or
604
perhaps we should have a new API.
605
532
606
533
This is topologically sorted.
607
534
"""
610
537
if not self.has_revision(revision_id):
611
538
raise errors.NoSuchRevision(self, revision_id)
612
539
w = self.get_inventory_weave()
613
candidates = w.get_ancestry(revision_id)
614
return [None] + candidates # self._eliminate_revisions_not_present(candidates)
540
return [None] + map(w.idx_to_name,
541
w.inclusions([w.lookup(revision_id)]))
615
542
616
543
@needs_read_lock
617
544
def print_file(self, file, revision_id):
625
552
# use inventory as it was in that revision
626
553
file_id = tree.inventory.path2id(file)
627
554
if not file_id:
628
# TODO: jam 20060427 Write a test for this code path
629
# it had a bug in it, and was raising the wrong
630
# exception.
631
raise errors.BzrError("%r is not present in revision %s" % (file, revision_id))
555
raise BzrError("%r is not present in revision %s" % (file, revno))
556
try:
557
revno = self.revision_id_to_revno(revision_id)
558
except errors.NoSuchRevision:
559
# TODO: This should not be BzrError,
560
# but NoSuchFile doesn't fit either
561
raise BzrError('%r is not present in revision %s'
562
% (file, revision_id))
563
else:
564
raise BzrError('%r is not present in revision %s'
565
% (file, revno))
632
566
tree.print_file(file_id)
633
567
634
568
def get_transaction(self):
635
569
return self.control_files.get_transaction()
636
570
637
def revision_parents(self, revid):
638
return self.get_inventory_weave().parent_names(revid)
639
640
@needs_write_lock
641
def set_make_working_trees(self, new_value):
642
"""Set the policy flag for making working trees when creating branches.
643
644
This only applies to branches that use this repository.
645
646
The default is 'True'.
647
:param new_value: True to restore the default, False to disable making
648
working trees.
649
"""
650
raise NotImplementedError(self.set_make_working_trees)
651
652
def make_working_trees(self):
653
"""Returns the policy for making working trees on new branches."""
654
raise NotImplementedError(self.make_working_trees)
655
656
571
@needs_write_lock
657
572
def sign_revision(self, revision_id, gpg_strategy):
658
573
plaintext = Testament.from_revision(self, revision_id).as_short_text()
659
574
self.store_revision_signature(gpg_strategy, plaintext, revision_id)
660
575
661
@needs_read_lock
662
def has_signature_for_revision_id(self, revision_id):
663
"""Query for a revision signature for revision_id in the repository."""
664
return self._revision_store.has_signature(revision_id,
665
self.get_transaction())
666
667
@needs_read_lock
668
def get_signature_text(self, revision_id):
669
"""Return the text for a signature."""
670
return self._revision_store.get_signature_text(revision_id,
671
self.get_transaction())
672
673
@needs_read_lock
674
def check(self, revision_ids):
675
"""Check consistency of all history of given revision_ids.
676
677
Different repository implementations should override _check().
678
679
:param revision_ids: A non-empty list of revision_ids whose ancestry
680
will be checked. Typically the last revision_id of a branch.
681
"""
682
if not revision_ids:
683
raise ValueError("revision_ids must be non-empty in %s.check"
684
% (self,))
685
return self._check(revision_ids)
686
687
def _check(self, revision_ids):
688
result = check.Check(self)
689
result.check()
690
return result
691
692
def _warn_if_deprecated(self):
693
global _deprecation_warning_done
694
if _deprecation_warning_done:
695
return
696
_deprecation_warning_done = True
697
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
698
% (self._format, self.bzrdir.transport.base))
699
700
701
class AllInOneRepository(Repository):
702
"""Legacy support - the repository behaviour for all-in-one branches."""
703
704
def __init__(self, _format, a_bzrdir, _revision_store, control_store, text_store):
705
# we reuse one control files instance.
706
dir_mode = a_bzrdir._control_files._dir_mode
707
file_mode = a_bzrdir._control_files._file_mode
708
709
def get_store(name, compressed=True, prefixed=False):
710
# FIXME: This approach of assuming stores are all entirely compressed
711
# or entirely uncompressed is tidy, but breaks upgrade from
712
# some existing branches where there's a mixture; we probably
713
# still want the option to look for both.
714
relpath = a_bzrdir._control_files._escape(name)
715
store = TextStore(a_bzrdir._control_files._transport.clone(relpath),
716
prefixed=prefixed, compressed=compressed,
717
dir_mode=dir_mode,
718
file_mode=file_mode)
719
#if self._transport.should_cache():
720
# cache_path = os.path.join(self.cache_root, name)
721
# os.mkdir(cache_path)
722
# store = bzrlib.store.CachedStore(store, cache_path)
723
return store
724
725
# not broken out yet because the controlweaves|inventory_store
726
# and text_store | weave_store bits are still different.
727
if isinstance(_format, RepositoryFormat4):
728
# cannot remove these - there is still no consistent api
729
# which allows access to this old info.
730
self.inventory_store = get_store('inventory-store')
731
text_store = get_store('text-store')
732
super(AllInOneRepository, self).__init__(_format, a_bzrdir, a_bzrdir._control_files, _revision_store, control_store, text_store)
733
734
@needs_read_lock
735
def is_shared(self):
736
"""AllInOne repositories cannot be shared."""
737
return False
738
739
@needs_write_lock
740
def set_make_working_trees(self, new_value):
741
"""Set the policy flag for making working trees when creating branches.
742
743
This only applies to branches that use this repository.
744
745
The default is 'True'.
746
:param new_value: True to restore the default, False to disable making
747
working trees.
748
"""
749
raise NotImplementedError(self.set_make_working_trees)
750
751
def make_working_trees(self):
752
"""Returns the policy for making working trees on new branches."""
753
return True
754
755
756
def install_revision(repository, rev, revision_tree):
757
"""Install all revision data into a repository."""
758
present_parents = []
759
parent_trees = {}
760
for p_id in rev.parent_ids:
761
if repository.has_revision(p_id):
762
present_parents.append(p_id)
763
parent_trees[p_id] = repository.revision_tree(p_id)
764
else:
765
parent_trees[p_id] = repository.revision_tree(None)
766
767
inv = revision_tree.inventory
768
769
# backwards compatability hack: skip the root id.
770
entries = inv.iter_entries()
771
entries.next()
772
# Add the texts that are not already present
773
for path, ie in entries:
774
w = repository.weave_store.get_weave_or_empty(ie.file_id,
775
repository.get_transaction())
776
if ie.revision not in w:
777
text_parents = []
778
# FIXME: TODO: The following loop *may* be overlapping/duplicate
779
# with InventoryEntry.find_previous_heads(). if it is, then there
780
# is a latent bug here where the parents may have ancestors of each
781
# other. RBC, AB
782
for revision, tree in parent_trees.iteritems():
783
if ie.file_id not in tree:
784
continue
785
parent_id = tree.inventory[ie.file_id].revision
786
if parent_id in text_parents:
787
continue
788
text_parents.append(parent_id)
789
790
vfile = repository.weave_store.get_weave_or_empty(ie.file_id,
791
repository.get_transaction())
792
lines = revision_tree.get_file(ie.file_id).readlines()
793
vfile.add_lines(rev.revision_id, text_parents, lines)
794
try:
795
# install the inventory
796
repository.add_inventory(rev.revision_id, inv, present_parents)
797
except errors.RevisionAlreadyPresent:
798
pass
799
repository.add_revision(rev.revision_id, rev, inv)
800
801
802
class MetaDirRepository(Repository):
803
"""Repositories in the new meta-dir layout."""
804
805
def __init__(self, _format, a_bzrdir, control_files, _revision_store, control_store, text_store):
806
super(MetaDirRepository, self).__init__(_format,
807
a_bzrdir,
808
control_files,
809
_revision_store,
810
control_store,
811
text_store)
812
dir_mode = self.control_files._dir_mode
813
file_mode = self.control_files._file_mode
814
815
@needs_read_lock
816
def is_shared(self):
817
"""Return True if this repository is flagged as a shared repository."""
818
return self.control_files._transport.has('shared-storage')
819
820
@needs_write_lock
821
def set_make_working_trees(self, new_value):
822
"""Set the policy flag for making working trees when creating branches.
823
824
This only applies to branches that use this repository.
825
826
The default is 'True'.
827
:param new_value: True to restore the default, False to disable making
828
working trees.
829
"""
830
if new_value:
831
try:
832
self.control_files._transport.delete('no-working-trees')
833
except errors.NoSuchFile:
834
pass
835
else:
836
self.control_files.put_utf8('no-working-trees', '')
837
838
def make_working_trees(self):
839
"""Returns the policy for making working trees on new branches."""
840
return not self.control_files._transport.has('no-working-trees')
841
842
843
class KnitRepository(MetaDirRepository):
844
"""Knit format repository."""
845
846
def _warn_if_deprecated(self):
847
# This class isn't deprecated
848
pass
849
850
def _inventory_add_lines(self, inv_vf, revid, parents, lines):
851
inv_vf.add_lines_with_ghosts(revid, parents, lines)
852
853
@needs_read_lock
854
def _all_revision_ids(self):
855
"""See Repository.all_revision_ids()."""
856
# Knits get the revision graph from the index of the revision knit, so
857
# it's always possible even if they're on an unlistable transport.
858
return self._revision_store.all_revision_ids(self.get_transaction())
859
860
def fileid_involved_between_revs(self, from_revid, to_revid):
861
"""Find file_id(s) which are involved in the changes between revisions.
862
863
This determines the set of revisions which are involved, and then
864
finds all file ids affected by those revisions.
865
"""
866
vf = self._get_revision_vf()
867
from_set = set(vf.get_ancestry(from_revid))
868
to_set = set(vf.get_ancestry(to_revid))
869
changed = to_set.difference(from_set)
870
return self._fileid_involved_by_set(changed)
871
872
def fileid_involved(self, last_revid=None):
873
"""Find all file_ids modified in the ancestry of last_revid.
874
875
:param last_revid: If None, last_revision() will be used.
876
"""
877
if not last_revid:
878
changed = set(self.all_revision_ids())
879
else:
880
changed = set(self.get_ancestry(last_revid))
881
if None in changed:
882
changed.remove(None)
883
return self._fileid_involved_by_set(changed)
884
885
@needs_read_lock
886
def get_ancestry(self, revision_id):
887
"""Return a list of revision-ids integrated by a revision.
888
889
This is topologically sorted.
890
"""
891
if revision_id is None:
892
return [None]
893
vf = self._get_revision_vf()
894
try:
895
return [None] + vf.get_ancestry(revision_id)
896
except errors.RevisionNotPresent:
897
raise errors.NoSuchRevision(self, revision_id)
898
899
@needs_read_lock
900
def get_revision(self, revision_id):
901
"""Return the Revision object for a named revision"""
902
return self.get_revision_reconcile(revision_id)
903
904
@needs_read_lock
905
def get_revision_graph(self, revision_id=None):
906
"""Return a dictionary containing the revision graph.
907
908
:param revision_id: The revision_id to get a graph from. If None, then
909
the entire revision graph is returned. This is a deprecated mode of
910
operation and will be removed in the future.
911
:return: a dictionary of revision_id->revision_parents_list.
912
"""
913
# special case NULL_REVISION
914
if revision_id == NULL_REVISION:
915
return {}
916
weave = self._get_revision_vf()
917
entire_graph = weave.get_graph()
918
if revision_id is None:
919
return weave.get_graph()
920
elif revision_id not in weave:
921
raise errors.NoSuchRevision(self, revision_id)
922
else:
923
# add what can be reached from revision_id
924
result = {}
925
pending = set([revision_id])
926
while len(pending) > 0:
927
node = pending.pop()
928
result[node] = weave.get_parents(node)
929
for revision_id in result[node]:
930
if revision_id not in result:
931
pending.add(revision_id)
932
return result
933
934
@needs_read_lock
935
def get_revision_graph_with_ghosts(self, revision_ids=None):
936
"""Return a graph of the revisions with ghosts marked as applicable.
937
938
:param revision_ids: an iterable of revisions to graph or None for all.
939
:return: a Graph object with the graph reachable from revision_ids.
940
"""
941
result = Graph()
942
vf = self._get_revision_vf()
943
versions = set(vf.versions())
944
if not revision_ids:
945
pending = set(self.all_revision_ids())
946
required = set([])
947
else:
948
pending = set(revision_ids)
949
# special case NULL_REVISION
950
if NULL_REVISION in pending:
951
pending.remove(NULL_REVISION)
952
required = set(pending)
953
done = set([])
954
while len(pending):
955
revision_id = pending.pop()
956
if not revision_id in versions:
957
if revision_id in required:
958
raise errors.NoSuchRevision(self, revision_id)
959
# a ghost
960
result.add_ghost(revision_id)
961
# mark it as done so we don't try for it again.
962
done.add(revision_id)
963
continue
964
parent_ids = vf.get_parents_with_ghosts(revision_id)
965
for parent_id in parent_ids:
966
# is this queued or done ?
967
if (parent_id not in pending and
968
parent_id not in done):
969
# no, queue it.
970
pending.add(parent_id)
971
result.add_node(revision_id, parent_ids)
972
done.add(revision_id)
973
return result
974
975
def _get_revision_vf(self):
976
""":return: a versioned file containing the revisions."""
977
vf = self._revision_store.get_revision_file(self.get_transaction())
978
return vf
979
980
@needs_write_lock
981
def reconcile(self, other=None, thorough=False):
982
"""Reconcile this repository."""
983
from bzrlib.reconcile import KnitReconciler
984
reconciler = KnitReconciler(self, thorough=thorough)
985
reconciler.reconcile()
986
return reconciler
987
988
def revision_parents(self, revision_id):
989
return self._get_revision_vf().get_parents(revision_id)
990
991
576
992
577
class RepositoryFormat(object):
993
578
"""A repository format.
1019
604
_formats = {}
1020
605
"""The known formats."""
1021
606
1022
def __str__(self):
1023
return "<%s>" % self.__class__.__name__
1024
1025
607
@classmethod
1026
608
def find_format(klass, a_bzrdir):
1027
609
"""Return the format for the repository object in a_bzrdir."""
1032
614
except errors.NoSuchFile:
1033
615
raise errors.NoRepositoryPresent(a_bzrdir)
1034
616
except KeyError:
1035
raise errors.UnknownFormatError(format=format_string)
617
raise errors.UnknownFormatError(format_string)
1036
618
1037
def _get_control_store(self, repo_transport, control_files):
1038
"""Return the control store for this repository."""
1039
raise NotImplementedError(self._get_control_store)
1040
1041
619
@classmethod
1042
620
def get_default_format(klass):
1043
621
"""Return the current default format."""
1051
629
"""
1052
630
raise NotImplementedError(self.get_format_string)
1053
631
1054
def get_format_description(self):
1055
"""Return the short description for this format."""
1056
raise NotImplementedError(self.get_format_description)
1057
1058
def _get_revision_store(self, repo_transport, control_files):
1059
"""Return the revision store object for this a_bzrdir."""
1060
raise NotImplementedError(self._get_revision_store)
1061
1062
def _get_text_rev_store(self,
1063
transport,
1064
control_files,
1065
name,
1066
compressed=True,
1067
prefixed=False,
1068
serializer=None):
1069
"""Common logic for getting a revision store for a repository.
1070
1071
see self._get_revision_store for the subclass-overridable method to
1072
get the store for a repository.
1073
"""
1074
from bzrlib.store.revision.text import TextRevisionStore
1075
dir_mode = control_files._dir_mode
1076
file_mode = control_files._file_mode
1077
text_store =TextStore(transport.clone(name),
1078
prefixed=prefixed,
1079
compressed=compressed,
1080
dir_mode=dir_mode,
1081
file_mode=file_mode)
1082
_revision_store = TextRevisionStore(text_store, serializer)
1083
return _revision_store
1084
1085
def _get_versioned_file_store(self,
1086
name,
1087
transport,
1088
control_files,
1089
prefixed=True,
1090
versionedfile_class=WeaveFile,
1091
escaped=False):
1092
weave_transport = control_files._transport.clone(name)
1093
dir_mode = control_files._dir_mode
1094
file_mode = control_files._file_mode
1095
return VersionedFileStore(weave_transport, prefixed=prefixed,
1096
dir_mode=dir_mode,
1097
file_mode=file_mode,
1098
versionedfile_class=versionedfile_class,
1099
escaped=escaped)
1100
1101
def initialize(self, a_bzrdir, shared=False):
1102
"""Initialize a repository of this format in a_bzrdir.
1103
1104
:param a_bzrdir: The bzrdir to put the new repository in it.
1105
:param shared: The repository should be initialized as a sharable one.
1106
1107
This may raise UninitializableFormat if shared repository are not
1108
compatible the a_bzrdir.
1109
"""
632
def initialize(self, a_bzrdir, _internal=False):
633
"""Create a weave repository.
634
635
TODO: when creating split out bzr branch formats, move this to a common
636
base for Format5, Format6. or something like that.
637
"""
638
from bzrlib.weavefile import write_weave_v5
639
from bzrlib.weave import Weave
640
641
if not _internal:
642
# always initialized when the bzrdir is.
643
return Repository(None, branch_format=None, _format=self, a_bzrdir=a_bzrdir)
644
645
# Create an empty weave
646
sio = StringIO()
647
bzrlib.weavefile.write_weave_v5(Weave(), sio)
648
empty_weave = sio.getvalue()
649
650
mutter('creating repository in %s.', a_bzrdir.transport.base)
651
dirs = ['revision-store', 'weaves']
652
lock_file = 'branch-lock'
653
files = [('inventory.weave', StringIO(empty_weave)),
654
]
655
656
# FIXME: RBC 20060125 dont peek under the covers
657
# NB: no need to escape relative paths that are url safe.
658
control_files = LockableFiles(a_bzrdir.transport, 'branch-lock')
659
control_files.lock_write()
660
control_files._transport.mkdir_multi(dirs,
661
mode=control_files._dir_mode)
662
try:
663
for file, content in files:
664
control_files.put(file, content)
665
finally:
666
control_files.unlock()
667
return Repository(None, branch_format=None, _format=self, a_bzrdir=a_bzrdir)
1110
668
1111
669
def is_supported(self):
1112
670
"""Is this format supported?
1122
680
1123
681
_found is a private parameter, do not use it.
1124
682
"""
1125
raise NotImplementedError(self.open)
683
if not _found:
684
# we are being called directly and must probe.
685
raise NotImplementedError
686
return Repository(None, branch_format=None, _format=self, a_bzrdir=a_bzrdir)
1126
687
1127
688
@classmethod
1128
689
def register_format(klass, format):
1138
699
del klass._formats[format.get_format_string()]
1139
700
1140
701
1141
class PreSplitOutRepositoryFormat(RepositoryFormat):
1142
"""Base class for the pre split out repository formats."""
1143
1144
def initialize(self, a_bzrdir, shared=False, _internal=False):
1145
"""Create a weave repository.
1146
1147
TODO: when creating split out bzr branch formats, move this to a common
1148
base for Format5, Format6. or something like that.
1149
"""
1150
from bzrlib.weavefile import write_weave_v5
1151
from bzrlib.weave import Weave
1152
1153
if shared:
1154
raise errors.IncompatibleFormat(self, a_bzrdir._format)
1155
1156
if not _internal:
1157
# always initialized when the bzrdir is.
1158
return self.open(a_bzrdir, _found=True)
1159
1160
# Create an empty weave
1161
sio = StringIO()
1162
write_weave_v5(Weave(), sio)
1163
empty_weave = sio.getvalue()
1164
1165
mutter('creating repository in %s.', a_bzrdir.transport.base)
1166
dirs = ['revision-store', 'weaves']
1167
files = [('inventory.weave', StringIO(empty_weave)),
1168
]
1169
1170
# FIXME: RBC 20060125 don't peek under the covers
1171
# NB: no need to escape relative paths that are url safe.
1172
control_files = LockableFiles(a_bzrdir.transport, 'branch-lock',
1173
TransportLock)
1174
control_files.create_lock()
1175
control_files.lock_write()
1176
control_files._transport.mkdir_multi(dirs,
1177
mode=control_files._dir_mode)
1178
try:
1179
for file, content in files:
1180
control_files.put(file, content)
1181
finally:
1182
control_files.unlock()
1183
return self.open(a_bzrdir, _found=True)
1184
1185
def _get_control_store(self, repo_transport, control_files):
1186
"""Return the control store for this repository."""
1187
return self._get_versioned_file_store('',
1188
repo_transport,
1189
control_files,
1190
prefixed=False)
1191
1192
def _get_text_store(self, transport, control_files):
1193
"""Get a store for file texts for this format."""
1194
raise NotImplementedError(self._get_text_store)
1195
1196
def open(self, a_bzrdir, _found=False):
1197
"""See RepositoryFormat.open()."""
1198
if not _found:
1199
# we are being called directly and must probe.
1200
raise NotImplementedError
1201
1202
repo_transport = a_bzrdir.get_repository_transport(None)
1203
control_files = a_bzrdir._control_files
1204
text_store = self._get_text_store(repo_transport, control_files)
1205
control_store = self._get_control_store(repo_transport, control_files)
1206
_revision_store = self._get_revision_store(repo_transport, control_files)
1207
return AllInOneRepository(_format=self,
1208
a_bzrdir=a_bzrdir,
1209
_revision_store=_revision_store,
1210
control_store=control_store,
1211
text_store=text_store)
1212
1213
1214
class RepositoryFormat4(PreSplitOutRepositoryFormat):
702
class RepositoryFormat4(RepositoryFormat):
1215
703
"""Bzr repository format 4.
1216
704
1217
705
This repository format has:
1219
707
- TextStores for texts, inventories,revisions.
1220
708
1221
709
This format is deprecated: it indexes texts using a text id which is
1222
removed in format 5; initialization and write support for this format
710
removed in format 5; initializationa and write support for this format
1223
711
has been removed.
1224
712
"""
1225
713
1227
715
super(RepositoryFormat4, self).__init__()
1228
716
self._matchingbzrdir = bzrdir.BzrDirFormat4()
1229
717
1230
def get_format_description(self):
1231
"""See RepositoryFormat.get_format_description()."""
1232
return "Repository format 4"
1233
1234
def initialize(self, url, shared=False, _internal=False):
718
def initialize(self, url, _internal=False):
1235
719
"""Format 4 branches cannot be created."""
1236
720
raise errors.UninitializableFormat(self)
1237
721
1244
728
"""
1245
729
return False
1246
730
1247
def _get_control_store(self, repo_transport, control_files):
1248
"""Format 4 repositories have no formal control store at this point.
1249
1250
This will cause any control-file-needing apis to fail - this is desired.
1251
"""
1252
return None
1253
1254
def _get_revision_store(self, repo_transport, control_files):
1255
"""See RepositoryFormat._get_revision_store()."""
1256
from bzrlib.xml4 import serializer_v4
1257
return self._get_text_rev_store(repo_transport,
1258
control_files,
1259
'revision-store',
1260
serializer=serializer_v4)
1261
1262
def _get_text_store(self, transport, control_files):
1263
"""See RepositoryFormat._get_text_store()."""
1264
1265
1266
class RepositoryFormat5(PreSplitOutRepositoryFormat):
731
732
class RepositoryFormat5(RepositoryFormat):
1267
733
"""Bzr control format 5.
1268
734
1269
735
This repository format has:
1276
742
super(RepositoryFormat5, self).__init__()
1277
743
self._matchingbzrdir = bzrdir.BzrDirFormat5()
1278
744
1279
def get_format_description(self):
1280
"""See RepositoryFormat.get_format_description()."""
1281
return "Weave repository format 5"
1282
1283
def _get_revision_store(self, repo_transport, control_files):
1284
"""See RepositoryFormat._get_revision_store()."""
1285
"""Return the revision store object for this a_bzrdir."""
1286
return self._get_text_rev_store(repo_transport,
1287
control_files,
1288
'revision-store',
1289
compressed=False)
1290
1291
def _get_text_store(self, transport, control_files):
1292
"""See RepositoryFormat._get_text_store()."""
1293
return self._get_versioned_file_store('weaves', transport, control_files, prefixed=False)
1294
1295
1296
class RepositoryFormat6(PreSplitOutRepositoryFormat):
745
746
class RepositoryFormat6(RepositoryFormat):
1297
747
"""Bzr control format 6.
1298
748
1299
749
This repository format has:
1306
756
super(RepositoryFormat6, self).__init__()
1307
757
self._matchingbzrdir = bzrdir.BzrDirFormat6()
1308
758
1309
def get_format_description(self):
1310
"""See RepositoryFormat.get_format_description()."""
1311
return "Weave repository format 6"
1312
1313
def _get_revision_store(self, repo_transport, control_files):
1314
"""See RepositoryFormat._get_revision_store()."""
1315
return self._get_text_rev_store(repo_transport,
1316
control_files,
1317
'revision-store',
1318
compressed=False,
1319
prefixed=True)
1320
1321
def _get_text_store(self, transport, control_files):
1322
"""See RepositoryFormat._get_text_store()."""
1323
return self._get_versioned_file_store('weaves', transport, control_files)
1324
1325
1326
class MetaDirRepositoryFormat(RepositoryFormat):
1327
"""Common base class for the new repositories using the metadir layout."""
1328
1329
def __init__(self):
1330
super(MetaDirRepositoryFormat, self).__init__()
1331
self._matchingbzrdir = bzrdir.BzrDirMetaFormat1()
1332
1333
def _create_control_files(self, a_bzrdir):
1334
"""Create the required files and the initial control_files object."""
1335
# FIXME: RBC 20060125 don't peek under the covers
759
760
class RepositoryFormat7(RepositoryFormat):
761
"""Bzr repository 7.
762
763
This repository format has:
764
- weaves for file texts and inventory
765
- hash subdirectory based stores.
766
- TextStores for revisions and signatures.
767
- a format marker of its own
768
"""
769
770
def get_format_string(self):
771
"""See RepositoryFormat.get_format_string()."""
772
return "Bazaar-NG Repository format 7"
773
774
def initialize(self, a_bzrdir):
775
"""Create a weave repository.
776
"""
777
from bzrlib.weavefile import write_weave_v5
778
from bzrlib.weave import Weave
779
780
# Create an empty weave
781
sio = StringIO()
782
bzrlib.weavefile.write_weave_v5(Weave(), sio)
783
empty_weave = sio.getvalue()
784
785
mutter('creating repository in %s.', a_bzrdir.transport.base)
786
dirs = ['revision-store', 'weaves']
787
files = [('inventory.weave', StringIO(empty_weave)),
788
]
789
utf8_files = [('format', self.get_format_string())]
790
791
# FIXME: RBC 20060125 dont peek under the covers
1336
792
# NB: no need to escape relative paths that are url safe.
793
lock_file = 'lock'
1337
794
repository_transport = a_bzrdir.get_repository_transport(self)
1338
control_files = LockableFiles(repository_transport, 'lock', LockDir)
1339
control_files.create_lock()
1340
return control_files
1341
1342
def _upload_blank_content(self, a_bzrdir, dirs, files, utf8_files, shared):
1343
"""Upload the initial blank content."""
1344
control_files = self._create_control_files(a_bzrdir)
795
repository_transport.put(lock_file, StringIO()) # TODO get the file mode from the bzrdir lock files., mode=file_mode)
796
control_files = LockableFiles(repository_transport, 'lock')
1345
797
control_files.lock_write()
798
control_files._transport.mkdir_multi(dirs,
799
mode=control_files._dir_mode)
1346
800
try:
1347
control_files._transport.mkdir_multi(dirs,
1348
mode=control_files._dir_mode)
1349
801
for file, content in files:
1350
802
control_files.put(file, content)
1351
803
for file, content in utf8_files:
1352
804
control_files.put_utf8(file, content)
1353
if shared == True:
1354
control_files.put_utf8('shared-storage', '')
1355
805
finally:
1356
806
control_files.unlock()
1357
1358
1359
class RepositoryFormat7(MetaDirRepositoryFormat):
1360
"""Bzr repository 7.
1361
1362
This repository format has:
1363
- weaves for file texts and inventory
1364
- hash subdirectory based stores.
1365
- TextStores for revisions and signatures.
1366
- a format marker of its own
1367
- an optional 'shared-storage' flag
1368
- an optional 'no-working-trees' flag
1369
"""
1370
1371
def _get_control_store(self, repo_transport, control_files):
1372
"""Return the control store for this repository."""
1373
return self._get_versioned_file_store('',
1374
repo_transport,
1375
control_files,
1376
prefixed=False)
1377
1378
def get_format_string(self):
1379
"""See RepositoryFormat.get_format_string()."""
1380
return "Bazaar-NG Repository format 7"
1381
1382
def get_format_description(self):
1383
"""See RepositoryFormat.get_format_description()."""
1384
return "Weave repository format 7"
1385
1386
def _get_revision_store(self, repo_transport, control_files):
1387
"""See RepositoryFormat._get_revision_store()."""
1388
return self._get_text_rev_store(repo_transport,
1389
control_files,
1390
'revision-store',
1391
compressed=False,
1392
prefixed=True,
1393
)
1394
1395
def _get_text_store(self, transport, control_files):
1396
"""See RepositoryFormat._get_text_store()."""
1397
return self._get_versioned_file_store('weaves',
1398
transport,
1399
control_files)
1400
1401
def initialize(self, a_bzrdir, shared=False):
1402
"""Create a weave repository.
1403
1404
:param shared: If true the repository will be initialized as a shared
1405
repository.
1406
"""
1407
from bzrlib.weavefile import write_weave_v5
1408
from bzrlib.weave import Weave
1409
1410
# Create an empty weave
1411
sio = StringIO()
1412
write_weave_v5(Weave(), sio)
1413
empty_weave = sio.getvalue()
1414
1415
mutter('creating repository in %s.', a_bzrdir.transport.base)
1416
dirs = ['revision-store', 'weaves']
1417
files = [('inventory.weave', StringIO(empty_weave)),
1418
]
1419
utf8_files = [('format', self.get_format_string())]
1420
1421
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1422
return self.open(a_bzrdir=a_bzrdir, _found=True)
1423
1424
def open(self, a_bzrdir, _found=False, _override_transport=None):
1425
"""See RepositoryFormat.open().
1426
1427
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1428
repository at a slightly different url
1429
than normal. I.e. during 'upgrade'.
1430
"""
1431
if not _found:
1432
format = RepositoryFormat.find_format(a_bzrdir)
1433
assert format.__class__ == self.__class__
1434
if _override_transport is not None:
1435
repo_transport = _override_transport
1436
else:
1437
repo_transport = a_bzrdir.get_repository_transport(None)
1438
control_files = LockableFiles(repo_transport, 'lock', LockDir)
1439
text_store = self._get_text_store(repo_transport, control_files)
1440
control_store = self._get_control_store(repo_transport, control_files)
1441
_revision_store = self._get_revision_store(repo_transport, control_files)
1442
return MetaDirRepository(_format=self,
1443
a_bzrdir=a_bzrdir,
1444
control_files=control_files,
1445
_revision_store=_revision_store,
1446
control_store=control_store,
1447
text_store=text_store)
1448
1449
1450
class RepositoryFormatKnit1(MetaDirRepositoryFormat):
1451
"""Bzr repository knit format 1.
1452
1453
This repository format has:
1454
- knits for file texts and inventory
1455
- hash subdirectory based stores.
1456
- knits for revisions and signatures
1457
- TextStores for revisions and signatures.
1458
- a format marker of its own
1459
- an optional 'shared-storage' flag
1460
- an optional 'no-working-trees' flag
1461
- a LockDir lock
1462
1463
This format was introduced in bzr 0.8.
1464
"""
1465
1466
def _get_control_store(self, repo_transport, control_files):
1467
"""Return the control store for this repository."""
1468
return VersionedFileStore(
1469
repo_transport,
1470
prefixed=False,
1471
file_mode=control_files._file_mode,
1472
versionedfile_class=KnitVersionedFile,
1473
versionedfile_kwargs={'factory':KnitPlainFactory()},
1474
)
1475
1476
def get_format_string(self):
1477
"""See RepositoryFormat.get_format_string()."""
1478
return "Bazaar-NG Knit Repository Format 1"
1479
1480
def get_format_description(self):
1481
"""See RepositoryFormat.get_format_description()."""
1482
return "Knit repository format 1"
1483
1484
def _get_revision_store(self, repo_transport, control_files):
1485
"""See RepositoryFormat._get_revision_store()."""
1486
from bzrlib.store.revision.knit import KnitRevisionStore
1487
versioned_file_store = VersionedFileStore(
1488
repo_transport,
1489
file_mode=control_files._file_mode,
1490
prefixed=False,
1491
precious=True,
1492
versionedfile_class=KnitVersionedFile,
1493
versionedfile_kwargs={'delta':False, 'factory':KnitPlainFactory()},
1494
escaped=True,
1495
)
1496
return KnitRevisionStore(versioned_file_store)
1497
1498
def _get_text_store(self, transport, control_files):
1499
"""See RepositoryFormat._get_text_store()."""
1500
return self._get_versioned_file_store('knits',
1501
transport,
1502
control_files,
1503
versionedfile_class=KnitVersionedFile,
1504
escaped=True)
1505
1506
def initialize(self, a_bzrdir, shared=False):
1507
"""Create a knit format 1 repository.
1508
1509
:param a_bzrdir: bzrdir to contain the new repository; must already
1510
be initialized.
1511
:param shared: If true the repository will be initialized as a shared
1512
repository.
1513
"""
1514
mutter('creating repository in %s.', a_bzrdir.transport.base)
1515
dirs = ['revision-store', 'knits']
1516
files = []
1517
utf8_files = [('format', self.get_format_string())]
1518
1519
self._upload_blank_content(a_bzrdir, dirs, files, utf8_files, shared)
1520
repo_transport = a_bzrdir.get_repository_transport(None)
1521
control_files = LockableFiles(repo_transport, 'lock', LockDir)
1522
control_store = self._get_control_store(repo_transport, control_files)
1523
transaction = transactions.WriteTransaction()
1524
# trigger a write of the inventory store.
1525
control_store.get_weave_or_empty('inventory', transaction)
1526
_revision_store = self._get_revision_store(repo_transport, control_files)
1527
_revision_store.has_revision_id('A', transaction)
1528
_revision_store.get_signature_file(transaction)
1529
return self.open(a_bzrdir=a_bzrdir, _found=True)
1530
1531
def open(self, a_bzrdir, _found=False, _override_transport=None):
1532
"""See RepositoryFormat.open().
1533
1534
:param _override_transport: INTERNAL USE ONLY. Allows opening the
1535
repository at a slightly different url
1536
than normal. I.e. during 'upgrade'.
1537
"""
1538
if not _found:
1539
format = RepositoryFormat.find_format(a_bzrdir)
1540
assert format.__class__ == self.__class__
1541
if _override_transport is not None:
1542
repo_transport = _override_transport
1543
else:
1544
repo_transport = a_bzrdir.get_repository_transport(None)
1545
control_files = LockableFiles(repo_transport, 'lock', LockDir)
1546
text_store = self._get_text_store(repo_transport, control_files)
1547
control_store = self._get_control_store(repo_transport, control_files)
1548
_revision_store = self._get_revision_store(repo_transport, control_files)
1549
return KnitRepository(_format=self,
1550
a_bzrdir=a_bzrdir,
1551
control_files=control_files,
1552
_revision_store=_revision_store,
1553
control_store=control_store,
1554
text_store=text_store)
807
return Repository(None, branch_format=None, _format=self, a_bzrdir=a_bzrdir)
808
809
def __init__(self):
810
super(RepositoryFormat7, self).__init__()
811
self._matchingbzrdir = bzrdir.BzrDirMetaFormat1()
1555
812
1556
813
1557
814
# formats which have no format string are not discoverable
1558
815
# and not independently creatable, so are not registered.
1559
RepositoryFormat.register_format(RepositoryFormat7())
1560
_default_format = RepositoryFormatKnit1()
1561
RepositoryFormat.register_format(_default_format)
1562
RepositoryFormat.set_default_format(_default_format)
816
__default_format = RepositoryFormat7()
817
RepositoryFormat.register_format(__default_format)
818
RepositoryFormat.set_default_format(__default_format)
1563
819
_legacy_formats = [RepositoryFormat4(),
1564
820
RepositoryFormat5(),
1565
821
RepositoryFormat6()]
1566
822
1567
823
1568
class InterRepository(InterObject):
1569
"""This class represents operations taking place between two repositories.
1570
1571
Its instances have methods like copy_content and fetch, and contain
1572
references to the source and target repositories these operations can be
1573
carried out on.
1574
1575
Often we will provide convenience methods on 'repository' which carry out
1576
operations with another repository - they will always forward to
1577
InterRepository.get(other).method_name(parameters).
1578
"""
1579
1580
_optimisers = set()
1581
"""The available optimised InterRepository types."""
1582
1583
@needs_write_lock
1584
def copy_content(self, revision_id=None, basis=None):
1585
"""Make a complete copy of the content in self into destination.
1586
1587
This is a destructive operation! Do not use it on existing
1588
repositories.
1589
1590
:param revision_id: Only copy the content needed to construct
1591
revision_id and its parents.
1592
:param basis: Copy the needed data preferentially from basis.
1593
"""
1594
try:
1595
self.target.set_make_working_trees(self.source.make_working_trees())
1596
except NotImplementedError:
1597
pass
1598
# grab the basis available data
1599
if basis is not None:
1600
self.target.fetch(basis, revision_id=revision_id)
1601
# but don't bother fetching if we have the needed data now.
1602
if (revision_id not in (None, NULL_REVISION) and
1603
self.target.has_revision(revision_id)):
1604
return
1605
self.target.fetch(self.source, revision_id=revision_id)
1606
1607
@needs_write_lock
1608
def fetch(self, revision_id=None, pb=None):
1609
"""Fetch the content required to construct revision_id.
1610
1611
The content is copied from source to target.
1612
1613
:param revision_id: if None all content is copied, if NULL_REVISION no
1614
content is copied.
1615
:param pb: optional progress bar to use for progress reports. If not
1616
provided a default one will be created.
1617
1618
Returns the copied revision count and the failed revisions in a tuple:
1619
(copied, failures).
1620
"""
1621
from bzrlib.fetch import GenericRepoFetcher
1622
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1623
self.source, self.source._format, self.target, self.target._format)
1624
f = GenericRepoFetcher(to_repository=self.target,
1625
from_repository=self.source,
1626
last_revision=revision_id,
1627
pb=pb)
1628
return f.count_copied, f.failed_revisions
1629
1630
@needs_read_lock
1631
def missing_revision_ids(self, revision_id=None):
1632
"""Return the revision ids that source has that target does not.
1633
1634
These are returned in topological order.
1635
1636
:param revision_id: only return revision ids included by this
1637
revision_id.
1638
"""
1639
# generic, possibly worst case, slow code path.
1640
target_ids = set(self.target.all_revision_ids())
1641
if revision_id is not None:
1642
source_ids = self.source.get_ancestry(revision_id)
1643
assert source_ids[0] == None
1644
source_ids.pop(0)
1645
else:
1646
source_ids = self.source.all_revision_ids()
1647
result_set = set(source_ids).difference(target_ids)
1648
# this may look like a no-op: its not. It preserves the ordering
1649
# other_ids had while only returning the members from other_ids
1650
# that we've decided we need.
1651
return [rev_id for rev_id in source_ids if rev_id in result_set]
1652
1653
1654
class InterWeaveRepo(InterRepository):
1655
"""Optimised code paths between Weave based repositories."""
1656
1657
_matching_repo_format = RepositoryFormat7()
1658
"""Repository format for testing with."""
1659
1660
@staticmethod
1661
def is_compatible(source, target):
1662
"""Be compatible with known Weave formats.
1663
1664
We don't test for the stores being of specific types because that
1665
could lead to confusing results, and there is no need to be
1666
overly general.
1667
"""
1668
try:
1669
return (isinstance(source._format, (RepositoryFormat5,
1670
RepositoryFormat6,
1671
RepositoryFormat7)) and
1672
isinstance(target._format, (RepositoryFormat5,
1673
RepositoryFormat6,
1674
RepositoryFormat7)))
1675
except AttributeError:
1676
return False
1677
1678
@needs_write_lock
1679
def copy_content(self, revision_id=None, basis=None):
1680
"""See InterRepository.copy_content()."""
1681
# weave specific optimised path:
1682
if basis is not None:
1683
# copy the basis in, then fetch remaining data.
1684
basis.copy_content_into(self.target, revision_id)
1685
# the basis copy_content_into could miss-set this.
1686
try:
1687
self.target.set_make_working_trees(self.source.make_working_trees())
1688
except NotImplementedError:
1689
pass
1690
self.target.fetch(self.source, revision_id=revision_id)
1691
else:
1692
try:
1693
self.target.set_make_working_trees(self.source.make_working_trees())
1694
except NotImplementedError:
1695
pass
1696
# FIXME do not peek!
1697
if self.source.control_files._transport.listable():
1698
pb = ui.ui_factory.nested_progress_bar()
1699
try:
1700
self.target.weave_store.copy_all_ids(
1701
self.source.weave_store,
1702
pb=pb,
1703
from_transaction=self.source.get_transaction(),
1704
to_transaction=self.target.get_transaction())
1705
pb.update('copying inventory', 0, 1)
1706
self.target.control_weaves.copy_multi(
1707
self.source.control_weaves, ['inventory'],
1708
from_transaction=self.source.get_transaction(),
1709
to_transaction=self.target.get_transaction())
1710
self.target._revision_store.text_store.copy_all_ids(
1711
self.source._revision_store.text_store,
1712
pb=pb)
1713
finally:
1714
pb.finished()
1715
else:
1716
self.target.fetch(self.source, revision_id=revision_id)
1717
1718
@needs_write_lock
1719
def fetch(self, revision_id=None, pb=None):
1720
"""See InterRepository.fetch()."""
1721
from bzrlib.fetch import GenericRepoFetcher
1722
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1723
self.source, self.source._format, self.target, self.target._format)
1724
f = GenericRepoFetcher(to_repository=self.target,
1725
from_repository=self.source,
1726
last_revision=revision_id,
1727
pb=pb)
1728
return f.count_copied, f.failed_revisions
1729
1730
@needs_read_lock
1731
def missing_revision_ids(self, revision_id=None):
1732
"""See InterRepository.missing_revision_ids()."""
1733
# we want all revisions to satisfy revision_id in source.
1734
# but we don't want to stat every file here and there.
1735
# we want then, all revisions other needs to satisfy revision_id
1736
# checked, but not those that we have locally.
1737
# so the first thing is to get a subset of the revisions to
1738
# satisfy revision_id in source, and then eliminate those that
1739
# we do already have.
1740
# this is slow on high latency connection to self, but as as this
1741
# disk format scales terribly for push anyway due to rewriting
1742
# inventory.weave, this is considered acceptable.
1743
# - RBC 20060209
1744
if revision_id is not None:
1745
source_ids = self.source.get_ancestry(revision_id)
1746
assert source_ids[0] == None
1747
source_ids.pop(0)
1748
else:
1749
source_ids = self.source._all_possible_ids()
1750
source_ids_set = set(source_ids)
1751
# source_ids is the worst possible case we may need to pull.
1752
# now we want to filter source_ids against what we actually
1753
# have in target, but don't try to check for existence where we know
1754
# we do not have a revision as that would be pointless.
1755
target_ids = set(self.target._all_possible_ids())
1756
possibly_present_revisions = target_ids.intersection(source_ids_set)
1757
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1758
required_revisions = source_ids_set.difference(actually_present_revisions)
1759
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1760
if revision_id is not None:
1761
# we used get_ancestry to determine source_ids then we are assured all
1762
# revisions referenced are present as they are installed in topological order.
1763
# and the tip revision was validated by get_ancestry.
1764
return required_topo_revisions
1765
else:
1766
# if we just grabbed the possibly available ids, then
1767
# we only have an estimate of whats available and need to validate
1768
# that against the revision records.
1769
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1770
1771
1772
class InterKnitRepo(InterRepository):
1773
"""Optimised code paths between Knit based repositories."""
1774
1775
_matching_repo_format = RepositoryFormatKnit1()
1776
"""Repository format for testing with."""
1777
1778
@staticmethod
1779
def is_compatible(source, target):
1780
"""Be compatible with known Knit formats.
1781
1782
We don't test for the stores being of specific types because that
1783
could lead to confusing results, and there is no need to be
1784
overly general.
1785
"""
1786
try:
1787
return (isinstance(source._format, (RepositoryFormatKnit1)) and
1788
isinstance(target._format, (RepositoryFormatKnit1)))
1789
except AttributeError:
1790
return False
1791
1792
@needs_write_lock
1793
def fetch(self, revision_id=None, pb=None):
1794
"""See InterRepository.fetch()."""
1795
from bzrlib.fetch import KnitRepoFetcher
1796
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
1797
self.source, self.source._format, self.target, self.target._format)
1798
f = KnitRepoFetcher(to_repository=self.target,
1799
from_repository=self.source,
1800
last_revision=revision_id,
1801
pb=pb)
1802
return f.count_copied, f.failed_revisions
1803
1804
@needs_read_lock
1805
def missing_revision_ids(self, revision_id=None):
1806
"""See InterRepository.missing_revision_ids()."""
1807
if revision_id is not None:
1808
source_ids = self.source.get_ancestry(revision_id)
1809
assert source_ids[0] == None
1810
source_ids.pop(0)
1811
else:
1812
source_ids = self.source._all_possible_ids()
1813
source_ids_set = set(source_ids)
1814
# source_ids is the worst possible case we may need to pull.
1815
# now we want to filter source_ids against what we actually
1816
# have in target, but don't try to check for existence where we know
1817
# we do not have a revision as that would be pointless.
1818
target_ids = set(self.target._all_possible_ids())
1819
possibly_present_revisions = target_ids.intersection(source_ids_set)
1820
actually_present_revisions = set(self.target._eliminate_revisions_not_present(possibly_present_revisions))
1821
required_revisions = source_ids_set.difference(actually_present_revisions)
1822
required_topo_revisions = [rev_id for rev_id in source_ids if rev_id in required_revisions]
1823
if revision_id is not None:
1824
# we used get_ancestry to determine source_ids then we are assured all
1825
# revisions referenced are present as they are installed in topological order.
1826
# and the tip revision was validated by get_ancestry.
1827
return required_topo_revisions
1828
else:
1829
# if we just grabbed the possibly available ids, then
1830
# we only have an estimate of whats available and need to validate
1831
# that against the revision records.
1832
return self.source._eliminate_revisions_not_present(required_topo_revisions)
1833
1834
InterRepository.register_optimiser(InterWeaveRepo)
1835
InterRepository.register_optimiser(InterKnitRepo)
1836
824
# TODO: jam 20060108 Create a new branch format, and as part of upgrade
825
# make sure that ancestry.weave is deleted (it is never used, but
826
# used to be created)
1837
827
1838
828
class RepositoryTestProviderAdapter(object):
1839
829
"""A tool to generate a suite testing multiple repository formats at once.
1863
853
new_test.id = make_new_test_id()
1864
854
result.addTest(new_test)
1865
855
return result
1866
1867
1868
class InterRepositoryTestProviderAdapter(object):
1869
"""A tool to generate a suite testing multiple inter repository formats.
1870
1871
This is done by copying the test once for each interrepo provider and injecting
1872
the transport_server, transport_readonly_server, repository_format and
1873
repository_to_format classes into each copy.
1874
Each copy is also given a new id() to make it easy to identify.
1875
"""
1876
1877
def __init__(self, transport_server, transport_readonly_server, formats):
1878
self._transport_server = transport_server
1879
self._transport_readonly_server = transport_readonly_server
1880
self._formats = formats
1881
1882
def adapt(self, test):
1883
result = TestSuite()
1884
for interrepo_class, repository_format, repository_format_to in self._formats:
1885
new_test = deepcopy(test)
1886
new_test.transport_server = self._transport_server
1887
new_test.transport_readonly_server = self._transport_readonly_server
1888
new_test.interrepo_class = interrepo_class
1889
new_test.repository_format = repository_format
1890
new_test.repository_format_to = repository_format_to
1891
def make_new_test_id():
1892
new_id = "%s(%s)" % (new_test.id(), interrepo_class.__name__)
1893
return lambda: new_id
1894
new_test.id = make_new_test_id()
1895
result.addTest(new_test)
1896
return result
1897
1898
@staticmethod
1899
def default_test_list():
1900
"""Generate the default list of interrepo permutations to test."""
1901
result = []
1902
# test the default InterRepository between format 6 and the current
1903
# default format.
1904
# XXX: robertc 20060220 reinstate this when there are two supported
1905
# formats which do not have an optimal code path between them.
1906
result.append((InterRepository,
1907
RepositoryFormat6(),
1908
RepositoryFormatKnit1()))
1909
for optimiser in InterRepository._optimisers:
1910
result.append((optimiser,
1911
optimiser._matching_repo_format,
1912
optimiser._matching_repo_format
1913
))
1914
# if there are specific combinations we want to use, we can add them
1915
# here.
1916
return result
1917
1918
1919
class CopyConverter(object):
1920
"""A repository conversion tool which just performs a copy of the content.
1921
1922
This is slow but quite reliable.
1923
"""
1924
1925
def __init__(self, target_format):
1926
"""Create a CopyConverter.
1927
1928
:param target_format: The format the resulting repository should be.
1929
"""
1930
self.target_format = target_format
1931
1932
def convert(self, repo, pb):
1933
"""Perform the conversion of to_convert, giving feedback via pb.
1934
1935
:param to_convert: The disk object to convert.
1936
:param pb: a progress bar to use for progress information.
1937
"""
1938
self.pb = pb
1939
self.count = 0
1940
self.total = 4
1941
# this is only useful with metadir layouts - separated repo content.
1942
# trigger an assertion if not such
1943
repo._format.get_format_string()
1944
self.repo_dir = repo.bzrdir
1945
self.step('Moving repository to repository.backup')
1946
self.repo_dir.transport.move('repository', 'repository.backup')
1947
backup_transport = self.repo_dir.transport.clone('repository.backup')
1948
self.source_repo = repo._format.open(self.repo_dir,
1949
_found=True,
1950
_override_transport=backup_transport)
1951
self.step('Creating new repository')
1952
converted = self.target_format.initialize(self.repo_dir,
1953
self.source_repo.is_shared())
1954
converted.lock_write()
1955
try:
1956
self.step('Copying content into repository.')
1957
self.source_repo.copy_content_into(converted)
1958
finally:
1959
converted.unlock()
1960
self.step('Deleting old repository content.')
1961
self.repo_dir.transport.delete_tree('repository.backup')
1962
self.pb.note('repository converted')
1963
1964
def step(self, message):
1965
"""Update the pb by a step."""
1966
self.count +=1
1967
self.pb.update(message, self.count, self.total)
1968
1969
1970
class CommitBuilder(object):
1971
"""Provides an interface to build up a commit.
1972
1973
This allows describing a tree to be committed without needing to
1974
know the internals of the format of the repository.
1975
"""
1976
1977
record_root_entry = False
1978
def __init__(self, repository, parents, config, timestamp=None,
1979
timezone=None, committer=None, revprops=None,
1980
revision_id=None):
1981
"""Initiate a CommitBuilder.
1982
1983
:param repository: Repository to commit to.
1984
:param parents: Revision ids of the parents of the new revision.
1985
:param config: Configuration to use.
1986
:param timestamp: Optional timestamp recorded for commit.
1987
:param timezone: Optional timezone for timestamp.
1988
:param committer: Optional committer to set for commit.
1989
:param revprops: Optional dictionary of revision properties.
1990
:param revision_id: Optional revision id.
1991
"""
1992
self._config = config
1993
1994
if committer is None:
1995
self._committer = self._config.username()
1996
else:
1997
assert isinstance(committer, basestring), type(committer)
1998
self._committer = committer
1999
2000
self.new_inventory = Inventory(None)
2001
self._new_revision_id = revision_id
2002
self.parents = parents
2003
self.repository = repository
2004
2005
self._revprops = {}
2006
if revprops is not None:
2007
self._revprops.update(revprops)
2008
2009
if timestamp is None:
2010
timestamp = time.time()
2011
# Restrict resolution to 1ms
2012
self._timestamp = round(timestamp, 3)
2013
2014
if timezone is None:
2015
self._timezone = local_time_offset()
2016
else:
2017
self._timezone = int(timezone)
2018
2019
self._generate_revision_if_needed()
2020
2021
def commit(self, message):
2022
"""Make the actual commit.
2023
2024
:return: The revision id of the recorded revision.
2025
"""
2026
rev = Revision(timestamp=self._timestamp,
2027
timezone=self._timezone,
2028
committer=self._committer,
2029
message=message,
2030
inventory_sha1=self.inv_sha1,
2031
revision_id=self._new_revision_id,
2032
properties=self._revprops)
2033
rev.parent_ids = self.parents
2034
self.repository.add_revision(self._new_revision_id, rev,
2035
self.new_inventory, self._config)
2036
return self._new_revision_id
2037
2038
def finish_inventory(self):
2039
"""Tell the builder that the inventory is finished."""
2040
if self.new_inventory.root is None:
2041
symbol_versioning.warn('Root entry should be supplied to'
2042
' record_entry_contents, as of bzr 0.10.',
2043
DeprecationWarning, stacklevel=2)
2044
self.new_inventory.add(InventoryDirectory(ROOT_ID, '', None))
2045
self.new_inventory.revision_id = self._new_revision_id
2046
self.inv_sha1 = self.repository.add_inventory(
2047
self._new_revision_id,
2048
self.new_inventory,
2049
self.parents
2050
)
2051
2052
def _gen_revision_id(self):
2053
"""Return new revision-id."""
2054
s = '%s-%s-' % (self._config.user_email(),
2055
compact_date(self._timestamp))
2056
s += hexlify(rand_bytes(8))
2057
return s
2058
2059
def _generate_revision_if_needed(self):
2060
"""Create a revision id if None was supplied.
2061
2062
If the repository can not support user-specified revision ids
2063
they should override this function and raise UnsupportedOperation
2064
if _new_revision_id is not None.
2065
2066
:raises: UnsupportedOperation
2067
"""
2068
if self._new_revision_id is None:
2069
self._new_revision_id = self._gen_revision_id()
2070
2071
def record_entry_contents(self, ie, parent_invs, path, tree):
2072
"""Record the content of ie from tree into the commit if needed.
2073
2074
Side effect: sets ie.revision when unchanged
2075
2076
:param ie: An inventory entry present in the commit.
2077
:param parent_invs: The inventories of the parent revisions of the
2078
commit.
2079
:param path: The path the entry is at in the tree.
2080
:param tree: The tree which contains this entry and should be used to
2081
obtain content.
2082
"""
2083
if self.new_inventory.root is None and ie.parent_id is not None:
2084
symbol_versioning.warn('Root entry should be supplied to'
2085
' record_entry_contents, as of bzr 0.10.',
2086
DeprecationWarning, stacklevel=2)
2087
self.record_entry_contents(tree.inventory.root.copy(), parent_invs,
2088
'', tree)
2089
self.new_inventory.add(ie)
2090
2091
# ie.revision is always None if the InventoryEntry is considered
2092
# for committing. ie.snapshot will record the correct revision
2093
# which may be the sole parent if it is untouched.
2094
if ie.revision is not None:
2095
return
2096
2097
# In this revision format, root entries have no knit or weave
2098
if ie is self.new_inventory.root:
2099
if len(parent_invs):
2100
ie.revision = parent_invs[0].root.revision
2101
else:
2102
ie.revision = None
2103
return
2104
previous_entries = ie.find_previous_heads(
2105
parent_invs,
2106
self.repository.weave_store,
2107
self.repository.get_transaction())
2108
# we are creating a new revision for ie in the history store
2109
# and inventory.
2110
ie.snapshot(self._new_revision_id, path, previous_entries, tree, self)
2111
2112
def modified_directory(self, file_id, file_parents):
2113
"""Record the presence of a symbolic link.
2114
2115
:param file_id: The file_id of the link to record.
2116
:param file_parents: The per-file parent revision ids.
2117
"""
2118
self._add_text_to_weave(file_id, [], file_parents.keys())
2119
2120
def modified_file_text(self, file_id, file_parents,
2121
get_content_byte_lines, text_sha1=None,
2122
text_size=None):
2123
"""Record the text of file file_id
2124
2125
:param file_id: The file_id of the file to record the text of.
2126
:param file_parents: The per-file parent revision ids.
2127
:param get_content_byte_lines: A callable which will return the byte
2128
lines for the file.
2129
:param text_sha1: Optional SHA1 of the file contents.
2130
:param text_size: Optional size of the file contents.
2131
"""
2132
# mutter('storing text of file {%s} in revision {%s} into %r',
2133
# file_id, self._new_revision_id, self.repository.weave_store)
2134
# special case to avoid diffing on renames or
2135
# reparenting
2136
if (len(file_parents) == 1
2137
and text_sha1 == file_parents.values()[0].text_sha1
2138
and text_size == file_parents.values()[0].text_size):
2139
previous_ie = file_parents.values()[0]
2140
versionedfile = self.repository.weave_store.get_weave(file_id,
2141
self.repository.get_transaction())
2142
versionedfile.clone_text(self._new_revision_id,
2143
previous_ie.revision, file_parents.keys())
2144
return text_sha1, text_size
2145
else:
2146
new_lines = get_content_byte_lines()
2147
# TODO: Rather than invoking sha_strings here, _add_text_to_weave
2148
# should return the SHA1 and size
2149
self._add_text_to_weave(file_id, new_lines, file_parents.keys())
2150
return osutils.sha_strings(new_lines), \
2151
sum(map(len, new_lines))
2152
2153
def modified_link(self, file_id, file_parents, link_target):
2154
"""Record the presence of a symbolic link.
2155
2156
:param file_id: The file_id of the link to record.
2157
:param file_parents: The per-file parent revision ids.
2158
:param link_target: Target location of this link.
2159
"""
2160
self._add_text_to_weave(file_id, [], file_parents.keys())
2161
2162
def _add_text_to_weave(self, file_id, new_lines, parents):
2163
versionedfile = self.repository.weave_store.get_weave_or_empty(
2164
file_id, self.repository.get_transaction())
2165
versionedfile.add_lines(self._new_revision_id, parents, new_lines)
2166
versionedfile.clear_cache()
2167
2168
2169
class _CommitBuilder(CommitBuilder):
2170
"""Temporary class so old CommitBuilders are detected properly
2171
2172
Note: CommitBuilder works whether or not root entry is recorded.
2173
"""
2174
2175
record_root_entry = True
2176
2177
2178
_unescape_map = {
2179
'apos':"'",
2180
'quot':'"',
2181
'amp':'&',
2182
'lt':'<',
2183
'gt':'>'
2184
}
2185
2186
2187
def _unescaper(match, _map=_unescape_map):
2188
return _map[match.group(1)]
2189
2190
2191
_unescape_re = None
2192
2193
2194
def _unescape_xml(data):
2195
"""Unescape predefined XML entities in a string of data."""
2196
global _unescape_re
2197
if _unescape_re is None:
2198
_unescape_re = re.compile('\&([^;]*);')
2199
return _unescape_re.sub(_unescaper, data)
Loggerhead is a web-based interface for Breezy
Version: 2.0.1