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 then return to the inventories.
34
import bzrlib.errors as errors
35
from bzrlib.errors import InstallFailed
36
from bzrlib.progress import ProgressPhase
37
from bzrlib.revision import is_null, NULL_REVISION
38
from bzrlib.symbol_versioning import (deprecated_function,
41
from bzrlib.trace import mutter
44
from bzrlib.lazy_import import lazy_import
46
# TODO: Avoid repeatedly opening weaves so many times.
48
# XXX: This doesn't handle ghost (not present in branch) revisions at
49
# all yet. I'm not sure they really should be supported.
51
# NOTE: This doesn't copy revisions which may be present but not
52
# merged into the last revision. I'm not sure we want to do that.
54
# - get a list of revisions that need to be pulled in
55
# - for each one, pull in that revision file
56
# and get the inventory, and store the inventory with right
58
# - and get the ancestry, and store that with right parents too
59
# - and keep a note of all file ids and version seen
60
# - then go through all files; for each one get the weave,
61
# and add in all file versions
64
class RepoFetcher(object):
65
"""Pull revisions and texts from one repository to another.
68
if set, try to limit to the data this revision references.
71
count_copied -- number of revisions copied
73
This should not be used directly, it's essential a object to encapsulate
74
the logic in InterRepository.fetch().
77
def __init__(self, to_repository, from_repository, last_revision=None, pb=None,
79
"""Create a repo fetcher.
81
:param find_ghosts: If True search the entire history for ghosts.
84
self.failed_revisions = []
86
if to_repository.has_same_location(from_repository):
87
# repository.fetch should be taking care of this case.
88
raise errors.BzrError('RepoFetcher run '
89
'between two objects at the same location: '
90
'%r and %r' % (to_repository, from_repository))
91
self.to_repository = to_repository
92
self.from_repository = from_repository
93
# must not mutate self._last_revision as its potentially a shared instance
94
self._last_revision = last_revision
95
self.find_ghosts = find_ghosts
97
self.pb = bzrlib.ui.ui_factory.nested_progress_bar()
98
self.nested_pb = self.pb
101
self.nested_pb = None
102
self.from_repository.lock_read()
104
self.to_repository.lock_write()
106
self.to_repository.start_write_group()
110
self.to_repository.abort_write_group()
113
self.to_repository.commit_write_group()
116
if self.nested_pb is not None:
117
self.nested_pb.finished()
119
self.to_repository.unlock()
121
self.from_repository.unlock()
124
"""Primary worker function.
126
This initialises all the needed variables, and then fetches the
127
requested revisions, finally clearing the progress bar.
129
self.to_weaves = self.to_repository.weave_store
130
self.from_weaves = self.from_repository.weave_store
132
self.file_ids_names = {}
133
pp = ProgressPhase('Transferring', 4, self.pb)
136
search = self._revids_to_fetch()
139
if getattr(self, '_fetch_everything_for_search', None) is not None:
140
self._fetch_everything_for_search(search, pp)
142
# backward compatibility
143
self._fetch_everything_for_revisions(search.get_keys, pp)
147
def _fetch_everything_for_search(self, search, pp):
148
"""Fetch all data for the given set of revisions."""
149
# The first phase is "file". We pass the progress bar for it directly
150
# into item_keys_introduced_by, which has more information about how
151
# that phase is progressing than we do. Progress updates for the other
152
# phases are taken care of in this function.
153
# XXX: there should be a clear owner of the progress reporting. Perhaps
154
# item_keys_introduced_by should have a richer API than it does at the
155
# moment, so that it can feed the progress information back to this
158
pb = bzrlib.ui.ui_factory.nested_progress_bar()
160
revs = search.get_keys()
161
data_to_fetch = self.from_repository.item_keys_introduced_by(revs, pb)
162
for knit_kind, file_id, revisions in data_to_fetch:
163
if knit_kind != phase:
165
# Make a new progress bar for this phase
168
pb = bzrlib.ui.ui_factory.nested_progress_bar()
169
if knit_kind == "file":
170
self._fetch_weave_text(file_id, revisions)
171
elif knit_kind == "inventory":
172
# Before we process the inventory we generate the root
173
# texts (if necessary) so that the inventories references
175
self._generate_root_texts(revs)
176
# NB: This currently reopens the inventory weave in source;
177
# using a full get_data_stream instead would avoid this.
178
self._fetch_inventory_weave(revs, pb)
179
elif knit_kind == "signatures":
180
# Nothing to do here; this will be taken care of when
181
# _fetch_revision_texts happens.
183
elif knit_kind == "revisions":
184
self._fetch_revision_texts(revs, pb)
186
raise AssertionError("Unknown knit kind %r" % knit_kind)
190
self.count_copied += len(revs)
192
def _revids_to_fetch(self):
193
"""Determines the exact revisions needed from self.from_repository to
194
install self._last_revision in self.to_repository.
196
If no revisions need to be fetched, then this just returns None.
198
mutter('fetch up to rev {%s}', self._last_revision)
199
if self._last_revision is NULL_REVISION:
200
# explicit limit of no revisions needed
202
if (self._last_revision is not None and
203
self.to_repository.has_revision(self._last_revision)):
206
return self.to_repository.search_missing_revision_ids(
207
self.from_repository, self._last_revision,
208
find_ghosts=self.find_ghosts)
209
except errors.NoSuchRevision:
210
raise InstallFailed([self._last_revision])
212
def _fetch_weave_text(self, file_id, required_versions):
213
to_weave = self.to_weaves.get_weave_or_empty(file_id,
214
self.to_repository.get_transaction())
215
from_weave = self.from_weaves.get_weave(file_id,
216
self.from_repository.get_transaction())
217
# we fetch all the texts, because texts do
218
# not reference anything, and its cheap enough
219
to_weave.join(from_weave, version_ids=required_versions)
221
def _fetch_inventory_weave(self, revs, pb):
222
pb.update("fetch inventory", 0, 2)
223
to_weave = self.to_repository.get_inventory_weave()
224
child_pb = bzrlib.ui.ui_factory.nested_progress_bar()
226
# just merge, this is optimisable and its means we don't
227
# copy unreferenced data such as not-needed inventories.
228
pb.update("fetch inventory", 1, 3)
229
from_weave = self.from_repository.get_inventory_weave()
230
pb.update("fetch inventory", 2, 3)
231
# we fetch only the referenced inventories because we do not
232
# know for unselected inventories whether all their required
233
# texts are present in the other repository - it could be
235
to_weave.join(from_weave, pb=child_pb, msg='merge inventory',
240
def _generate_root_texts(self, revs):
241
"""This will be called by __fetch between fetching weave texts and
242
fetching the inventory weave.
244
Subclasses should override this if they need to generate root texts
245
after fetching weave texts.
250
class GenericRepoFetcher(RepoFetcher):
251
"""This is a generic repo to repo fetcher.
253
This makes minimal assumptions about repo layout and contents.
254
It triggers a reconciliation after fetching to ensure integrity.
257
def _fetch_revision_texts(self, revs, pb):
258
"""Fetch revision object texts"""
259
to_txn = self.to_transaction = self.to_repository.get_transaction()
262
to_store = self.to_repository._revision_store
264
pb.update('copying revisions', count, total)
266
sig_text = self.from_repository.get_signature_text(rev)
267
to_store.add_revision_signature_text(rev, sig_text, to_txn)
268
except errors.NoSuchRevision:
271
to_store.add_revision(self.from_repository.get_revision(rev),
274
# fixup inventory if needed:
275
# this is expensive because we have no inverse index to current ghosts.
276
# but on local disk its a few seconds and sftp push is already insane.
278
# FIXME: repository should inform if this is needed.
279
self.to_repository.reconcile()
282
class KnitRepoFetcher(RepoFetcher):
283
"""This is a knit format repository specific fetcher.
285
This differs from the GenericRepoFetcher by not doing a
286
reconciliation after copying, and using knit joining to
290
def _fetch_revision_texts(self, revs, pb):
291
# may need to be a InterRevisionStore call here.
292
from_transaction = self.from_repository.get_transaction()
293
to_transaction = self.to_repository.get_transaction()
294
to_sf = self.to_repository._revision_store.get_signature_file(
296
from_sf = self.from_repository._revision_store.get_signature_file(
298
to_sf.join(from_sf, version_ids=revs, ignore_missing=True)
299
to_rf = self.to_repository._revision_store.get_revision_file(
301
from_rf = self.from_repository._revision_store.get_revision_file(
303
to_rf.join(from_rf, version_ids=revs)
306
class Inter1and2Helper(object):
307
"""Helper for operations that convert data from model 1 and 2
309
This is for use by fetchers and converters.
312
def __init__(self, source, target):
315
:param source: The repository data comes from
316
:param target: The repository data goes to
321
def iter_rev_trees(self, revs):
322
"""Iterate through RevisionTrees efficiently.
324
Additionally, the inventory's revision_id is set if unset.
326
Trees are retrieved in batches of 100, and then yielded in the order
329
:param revs: A list of revision ids
331
# In case that revs is not a list.
334
for tree in self.source.revision_trees(revs[:100]):
335
if tree.inventory.revision_id is None:
336
tree.inventory.revision_id = tree.get_revision_id()
340
def generate_root_texts(self, revs):
341
"""Generate VersionedFiles for all root ids.
343
:param revs: the revisions to include
345
inventory_weave = self.source.get_inventory_weave()
348
to_store = self.target.weave_store
349
parent_map = self.source.get_graph().get_parent_map(revs)
350
for tree in self.iter_rev_trees(revs):
351
revision_id = tree.inventory.root.revision
352
root_id = tree.get_root_id()
353
parents = parent_map[revision_id]
354
if parents[0] == NULL_REVISION:
356
if root_id not in versionedfile:
357
versionedfile[root_id] = to_store.get_weave_or_empty(root_id,
358
self.target.get_transaction())
359
_, _, parent_texts[root_id] = versionedfile[root_id].add_lines(
360
revision_id, parents, [], parent_texts)
362
def regenerate_inventory(self, revs):
363
"""Generate a new inventory versionedfile in target, convertin data.
365
The inventory is retrieved from the source, (deserializing it), and
366
stored in the target (reserializing it in a different format).
367
:param revs: The revisions to include
369
for tree in self.iter_rev_trees(revs):
370
parents = tree.get_parent_ids()
371
self.target.add_inventory(tree.get_revision_id(), tree.inventory,
375
class Model1toKnit2Fetcher(GenericRepoFetcher):
376
"""Fetch from a Model1 repository into a Knit2 repository
378
def __init__(self, to_repository, from_repository, last_revision=None,
379
pb=None, find_ghosts=True):
380
self.helper = Inter1and2Helper(from_repository, to_repository)
381
GenericRepoFetcher.__init__(self, to_repository, from_repository,
382
last_revision, pb, find_ghosts)
384
def _generate_root_texts(self, revs):
385
self.helper.generate_root_texts(revs)
387
def _fetch_inventory_weave(self, revs, pb):
388
self.helper.regenerate_inventory(revs)
391
class Knit1to2Fetcher(KnitRepoFetcher):
392
"""Fetch from a Knit1 repository into a Knit2 repository"""
394
def __init__(self, to_repository, from_repository, last_revision=None,
395
pb=None, find_ghosts=True):
396
self.helper = Inter1and2Helper(from_repository, to_repository)
397
KnitRepoFetcher.__init__(self, to_repository, from_repository,
398
last_revision, pb, find_ghosts)
400
def _generate_root_texts(self, revs):
401
self.helper.generate_root_texts(revs)
403
def _fetch_inventory_weave(self, revs, pb):
404
self.helper.regenerate_inventory(revs)
407
class RemoteToOtherFetcher(GenericRepoFetcher):
409
def _fetch_everything_for_search(self, search, pp):
410
data_stream = self.from_repository.get_data_stream_for_search(search)
411
self.to_repository.insert_data_stream(data_stream)