1
# Copyright (C) 2005, 2006 Canonical Ltd
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
18
"""Copying of history from one branch to another.
20
The basic plan is that every branch knows the history of everything
21
that has merged into it. As the first step of a merge, pull, or
22
branch operation we copy history from the source into the destination
25
The copying is done in a slightly complicated order. We don't want to
26
add a revision to the store until everything it refers to is also
27
stored, so that if a revision is present we can totally recreate it.
28
However, we can't know what files are included in a revision until we
29
read its inventory. So we query the inventory store of the source for
30
the ids we need, and then pull those ids and finally actually join
35
import bzrlib.errors as errors
36
from bzrlib.errors import (InstallFailed,
38
from bzrlib.progress import ProgressPhase
39
from bzrlib.revision import is_null, NULL_REVISION
40
from bzrlib.symbol_versioning import (deprecated_function,
43
from bzrlib.trace import mutter
46
from bzrlib.lazy_import import lazy_import
48
# TODO: Avoid repeatedly opening weaves so many times.
50
# XXX: This doesn't handle ghost (not present in branch) revisions at
51
# all yet. I'm not sure they really should be supported.
53
# NOTE: This doesn't copy revisions which may be present but not
54
# merged into the last revision. I'm not sure we want to do that.
56
# - get a list of revisions that need to be pulled in
57
# - for each one, pull in that revision file
58
# and get the inventory, and store the inventory with right
60
# - and get the ancestry, and store that with right parents too
61
# - and keep a note of all file ids and version seen
62
# - then go through all files; for each one get the weave,
63
# and add in all file versions
66
class RepoFetcher(object):
67
"""Pull revisions and texts from one repository to another.
70
if set, try to limit to the data this revision references.
73
count_copied -- number of revisions copied
75
This should not be used directly, it's essential a object to encapsulate
76
the logic in InterRepository.fetch().
78
def __init__(self, to_repository, from_repository, last_revision=None, pb=None):
80
self.failed_revisions = []
82
if to_repository.has_same_location(from_repository):
83
# repository.fetch should be taking care of this case.
84
raise errors.BzrError('RepoFetcher run '
85
'between two objects at the same location: '
86
'%r and %r' % (to_repository, from_repository))
87
self.to_repository = to_repository
88
self.from_repository = from_repository
89
# must not mutate self._last_revision as its potentially a shared instance
90
self._last_revision = last_revision
92
self.pb = bzrlib.ui.ui_factory.nested_progress_bar()
93
self.nested_pb = self.pb
97
self.from_repository.lock_read()
99
self.to_repository.lock_write()
101
self.to_repository.start_write_group()
105
self.to_repository.abort_write_group()
108
self.to_repository.commit_write_group()
110
if self.nested_pb is not None:
111
self.nested_pb.finished()
112
self.to_repository.unlock()
114
self.from_repository.unlock()
117
"""Primary worker function.
119
This initialises all the needed variables, and then fetches the
120
requested revisions, finally clearing the progress bar.
122
self.to_weaves = self.to_repository.weave_store
123
self.from_weaves = self.from_repository.weave_store
125
self.file_ids_names = {}
126
pp = ProgressPhase('Fetch phase', 4, self.pb)
129
revs = self._revids_to_fetch()
130
self._fetch_everything_for_revisions(revs, pp)
134
def _fetch_everything_for_revisions(self, revs, pp):
135
"""Fetch all data for the given set of revisions."""
138
# The first phase is "file". We pass the progress bar for it directly
139
# into item_keys_introduced_by, which has more information about how
140
# that phase is progressing than we do. Progress updates for the other
141
# phases are taken care of in this function.
142
# XXX: there should be a clear owner of the progress reporting. Perhaps
143
# item_keys_introduced_by should have a richer API than it does at the
144
# moment, so that it can feed the progress information back to this
147
pb = bzrlib.ui.ui_factory.nested_progress_bar()
149
data_to_fetch = self.from_repository.item_keys_introduced_by(revs, pb)
150
for knit_kind, file_id, revisions in data_to_fetch:
151
if knit_kind != phase:
153
# Make a new progress bar for this phase
156
pb = bzrlib.ui.ui_factory.nested_progress_bar()
157
if knit_kind == "file":
158
self._fetch_weave_text(file_id, revisions)
159
elif knit_kind == "inventory":
161
# Once we've processed all the files, then we generate the root
162
# texts (if necessary), then we process the inventory. It's a
163
# bit distasteful to have knit_kind == "inventory" mean this,
164
# perhaps it should happen on the first non-"file" knit, in case
165
# it's not always inventory?
166
self._generate_root_texts(revs)
167
self._fetch_inventory_weave(revs, pb)
168
elif knit_kind == "signatures":
169
# Nothing to do here; this will be taken care of when
170
# _fetch_revision_texts happens.
172
elif knit_kind == "revisions":
173
self._fetch_revision_texts(revs, pb)
175
raise AssertionError("Unknown knit kind %r" % knit_kind)
179
self.count_copied += len(revs)
181
def _revids_to_fetch(self):
182
"""Determines the exact revisions needed from self.from_repository to
183
install self._last_revision in self.to_repository.
185
If no revisions need to be fetched, then this just returns None.
187
mutter('fetch up to rev {%s}', self._last_revision)
188
if self._last_revision is NULL_REVISION:
189
# explicit limit of no revisions needed
191
if (self._last_revision is not None and
192
self.to_repository.has_revision(self._last_revision)):
196
return self.to_repository.missing_revision_ids(self.from_repository,
198
except errors.NoSuchRevision:
199
raise InstallFailed([self._last_revision])
201
def _fetch_weave_text(self, file_id, required_versions):
202
to_weave = self.to_weaves.get_weave_or_empty(file_id,
203
self.to_repository.get_transaction())
204
from_weave = self.from_weaves.get_weave(file_id,
205
self.from_repository.get_transaction())
206
# we fetch all the texts, because texts do
207
# not reference anything, and its cheap enough
208
to_weave.join(from_weave, version_ids=required_versions)
209
# we don't need *all* of this data anymore, but we dont know
210
# what we do. This cache clearing will result in a new read
211
# of the knit data when we do the checkout, but probably we
212
# want to emit the needed data on the fly rather than at the
214
# the from weave should know not to cache data being joined,
215
# but its ok to ask it to clear.
216
from_weave.clear_cache()
217
to_weave.clear_cache()
219
def _fetch_inventory_weave(self, revs, pb):
220
pb.update("fetch inventory", 0, 2)
221
to_weave = self.to_repository.get_inventory_weave()
222
child_pb = bzrlib.ui.ui_factory.nested_progress_bar()
224
# just merge, this is optimisable and its means we don't
225
# copy unreferenced data such as not-needed inventories.
226
pb.update("fetch inventory", 1, 3)
227
from_weave = self.from_repository.get_inventory_weave()
228
pb.update("fetch inventory", 2, 3)
229
# we fetch only the referenced inventories because we do not
230
# know for unselected inventories whether all their required
231
# texts are present in the other repository - it could be
233
to_weave.join(from_weave, pb=child_pb, msg='merge inventory',
235
from_weave.clear_cache()
239
def _generate_root_texts(self, revs):
240
"""This will be called by __fetch between fetching weave texts and
241
fetching the inventory weave.
243
Subclasses should override this if they need to generate root texts
244
after fetching weave texts.
249
class GenericRepoFetcher(RepoFetcher):
250
"""This is a generic repo to repo fetcher.
252
This makes minimal assumptions about repo layout and contents.
253
It triggers a reconciliation after fetching to ensure integrity.
256
def _fetch_revision_texts(self, revs, pb):
257
"""Fetch revision object texts"""
258
to_txn = self.to_transaction = self.to_repository.get_transaction()
261
to_store = self.to_repository._revision_store
263
pb.update('copying revisions', count, total)
265
sig_text = self.from_repository.get_signature_text(rev)
266
to_store.add_revision_signature_text(rev, sig_text, to_txn)
267
except errors.NoSuchRevision:
270
to_store.add_revision(self.from_repository.get_revision(rev),
273
# fixup inventory if needed:
274
# this is expensive because we have no inverse index to current ghosts.
275
# but on local disk its a few seconds and sftp push is already insane.
277
# FIXME: repository should inform if this is needed.
278
self.to_repository.reconcile()
281
class KnitRepoFetcher(RepoFetcher):
282
"""This is a knit format repository specific fetcher.
284
This differs from the GenericRepoFetcher by not doing a
285
reconciliation after copying, and using knit joining to
289
def _fetch_revision_texts(self, revs, pb):
290
# may need to be a InterRevisionStore call here.
291
from_transaction = self.from_repository.get_transaction()
292
to_transaction = self.to_repository.get_transaction()
293
to_sf = self.to_repository._revision_store.get_signature_file(
295
from_sf = self.from_repository._revision_store.get_signature_file(
297
to_sf.join(from_sf, version_ids=revs, ignore_missing=True)
298
to_rf = self.to_repository._revision_store.get_revision_file(
300
from_rf = self.from_repository._revision_store.get_revision_file(
302
to_rf.join(from_rf, version_ids=revs)
305
class Inter1and2Helper(object):
306
"""Helper for operations that convert data from model 1 and 2
308
This is for use by fetchers and converters.
311
def __init__(self, source, target):
314
:param source: The repository data comes from
315
:param target: The repository data goes to
320
def iter_rev_trees(self, revs):
321
"""Iterate through RevisionTrees efficiently.
323
Additionally, the inventory's revision_id is set if unset.
325
Trees are retrieved in batches of 100, and then yielded in the order
328
:param revs: A list of revision ids
331
for tree in self.source.revision_trees(revs[:100]):
332
if tree.inventory.revision_id is None:
333
tree.inventory.revision_id = tree.get_revision_id()
337
def generate_root_texts(self, revs):
338
"""Generate VersionedFiles for all root ids.
340
:param revs: the revisions to include
342
inventory_weave = self.source.get_inventory_weave()
345
to_store = self.target.weave_store
346
for tree in self.iter_rev_trees(revs):
347
revision_id = tree.inventory.root.revision
348
root_id = tree.inventory.root.file_id
349
parents = inventory_weave.get_parents(revision_id)
350
if root_id not in versionedfile:
351
versionedfile[root_id] = to_store.get_weave_or_empty(root_id,
352
self.target.get_transaction())
353
_, _, parent_texts[root_id] = versionedfile[root_id].add_lines(
354
revision_id, parents, [], parent_texts)
356
def regenerate_inventory(self, revs):
357
"""Generate a new inventory versionedfile in target, convertin data.
359
The inventory is retrieved from the source, (deserializing it), and
360
stored in the target (reserializing it in a different format).
361
:param revs: The revisions to include
363
inventory_weave = self.source.get_inventory_weave()
364
for tree in self.iter_rev_trees(revs):
365
parents = inventory_weave.get_parents(tree.get_revision_id())
366
self.target.add_inventory(tree.get_revision_id(), tree.inventory,
370
class Model1toKnit2Fetcher(GenericRepoFetcher):
371
"""Fetch from a Model1 repository into a Knit2 repository
373
def __init__(self, to_repository, from_repository, last_revision=None,
375
self.helper = Inter1and2Helper(from_repository, to_repository)
376
GenericRepoFetcher.__init__(self, to_repository, from_repository,
379
def _generate_root_texts(self, revs):
380
self.helper.generate_root_texts(revs)
382
def _fetch_inventory_weave(self, revs, pb):
383
self.helper.regenerate_inventory(revs)
386
class Knit1to2Fetcher(KnitRepoFetcher):
387
"""Fetch from a Knit1 repository into a Knit2 repository"""
389
def __init__(self, to_repository, from_repository, last_revision=None,
391
self.helper = Inter1and2Helper(from_repository, to_repository)
392
KnitRepoFetcher.__init__(self, to_repository, from_repository,
395
def _generate_root_texts(self, revs):
396
self.helper.generate_root_texts(revs)
398
def _fetch_inventory_weave(self, revs, pb):
399
self.helper.regenerate_inventory(revs)