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
37
from bzrlib.graph import SearchResult
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().
79
def __init__(self, to_repository, from_repository, last_revision=None, pb=None,
81
"""Create a repo fetcher.
83
:param find_ghosts: If True search the entire history for ghosts.
86
self.failed_revisions = []
88
if to_repository.has_same_location(from_repository):
89
# repository.fetch should be taking care of this case.
90
raise errors.BzrError('RepoFetcher run '
91
'between two objects at the same location: '
92
'%r and %r' % (to_repository, from_repository))
93
self.to_repository = to_repository
94
self.from_repository = from_repository
95
# must not mutate self._last_revision as its potentially a shared instance
96
self._last_revision = last_revision
97
self.find_ghosts = find_ghosts
99
self.pb = bzrlib.ui.ui_factory.nested_progress_bar()
100
self.nested_pb = self.pb
103
self.nested_pb = None
104
self.from_repository.lock_read()
106
self.to_repository.lock_write()
108
self.to_repository.start_write_group()
112
self.to_repository.abort_write_group()
115
self.to_repository.commit_write_group()
117
if self.nested_pb is not None:
118
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
revs = self._revids_to_fetch().get_keys()
139
self._fetch_everything_for_revisions(revs, pp)
143
def _fetch_everything_for_revisions(self, revs, pp):
144
"""Fetch all data for the given set of revisions."""
145
# The first phase is "file". We pass the progress bar for it directly
146
# into item_keys_introduced_by, which has more information about how
147
# that phase is progressing than we do. Progress updates for the other
148
# phases are taken care of in this function.
149
# XXX: there should be a clear owner of the progress reporting. Perhaps
150
# item_keys_introduced_by should have a richer API than it does at the
151
# moment, so that it can feed the progress information back to this
154
pb = bzrlib.ui.ui_factory.nested_progress_bar()
156
data_to_fetch = self.from_repository.item_keys_introduced_by(revs, pb)
157
for knit_kind, file_id, revisions in data_to_fetch:
158
if knit_kind != phase:
160
# Make a new progress bar for this phase
163
pb = bzrlib.ui.ui_factory.nested_progress_bar()
164
if knit_kind == "file":
165
self._fetch_weave_text(file_id, revisions)
166
elif knit_kind == "inventory":
168
# Once we've processed all the files, then we generate the root
169
# texts (if necessary), then we process the inventory. It's a
170
# bit distasteful to have knit_kind == "inventory" mean this,
171
# perhaps it should happen on the first non-"file" knit, in case
172
# it's not always inventory?
173
self._generate_root_texts(revs)
174
self._fetch_inventory_weave(revs, pb)
175
elif knit_kind == "signatures":
176
# Nothing to do here; this will be taken care of when
177
# _fetch_revision_texts happens.
179
elif knit_kind == "revisions":
180
self._fetch_revision_texts(revs, pb)
182
raise AssertionError("Unknown knit kind %r" % knit_kind)
186
self.count_copied += len(revs)
188
def _revids_to_fetch(self):
189
"""Determines the exact revisions needed from self.from_repository to
190
install self._last_revision in self.to_repository.
192
If no revisions need to be fetched, then this just returns None.
194
mutter('fetch up to rev {%s}', self._last_revision)
195
if self._last_revision is NULL_REVISION:
196
# explicit limit of no revisions needed
197
return SearchResult(set(), set(), 0, set())
198
if (self._last_revision is not None and
199
self.to_repository.has_revision(self._last_revision)):
200
return SearchResult(set(), set(), 0, set())
202
return self.to_repository.search_missing_revision_ids(
203
self.from_repository, self._last_revision,
204
find_ghosts=self.find_ghosts)
205
except errors.NoSuchRevision:
206
raise InstallFailed([self._last_revision])
208
def _fetch_weave_text(self, file_id, required_versions):
209
to_weave = self.to_weaves.get_weave_or_empty(file_id,
210
self.to_repository.get_transaction())
211
from_weave = self.from_weaves.get_weave(file_id,
212
self.from_repository.get_transaction())
213
# we fetch all the texts, because texts do
214
# not reference anything, and its cheap enough
215
to_weave.join(from_weave, version_ids=required_versions)
216
# we don't need *all* of this data anymore, but we dont know
217
# what we do. This cache clearing will result in a new read
218
# of the knit data when we do the checkout, but probably we
219
# want to emit the needed data on the fly rather than at the
221
# the from weave should know not to cache data being joined,
222
# but its ok to ask it to clear.
223
from_weave.clear_cache()
224
to_weave.clear_cache()
226
def _fetch_inventory_weave(self, revs, pb):
227
pb.update("fetch inventory", 0, 2)
228
to_weave = self.to_repository.get_inventory_weave()
229
child_pb = bzrlib.ui.ui_factory.nested_progress_bar()
231
# just merge, this is optimisable and its means we don't
232
# copy unreferenced data such as not-needed inventories.
233
pb.update("fetch inventory", 1, 3)
234
from_weave = self.from_repository.get_inventory_weave()
235
pb.update("fetch inventory", 2, 3)
236
# we fetch only the referenced inventories because we do not
237
# know for unselected inventories whether all their required
238
# texts are present in the other repository - it could be
240
to_weave.join(from_weave, pb=child_pb, msg='merge inventory',
242
from_weave.clear_cache()
246
def _generate_root_texts(self, revs):
247
"""This will be called by __fetch between fetching weave texts and
248
fetching the inventory weave.
250
Subclasses should override this if they need to generate root texts
251
after fetching weave texts.
256
class GenericRepoFetcher(RepoFetcher):
257
"""This is a generic repo to repo fetcher.
259
This makes minimal assumptions about repo layout and contents.
260
It triggers a reconciliation after fetching to ensure integrity.
263
def _fetch_revision_texts(self, revs, pb):
264
"""Fetch revision object texts"""
265
to_txn = self.to_transaction = self.to_repository.get_transaction()
268
to_store = self.to_repository._revision_store
270
pb.update('copying revisions', count, total)
272
sig_text = self.from_repository.get_signature_text(rev)
273
to_store.add_revision_signature_text(rev, sig_text, to_txn)
274
except errors.NoSuchRevision:
277
to_store.add_revision(self.from_repository.get_revision(rev),
280
# fixup inventory if needed:
281
# this is expensive because we have no inverse index to current ghosts.
282
# but on local disk its a few seconds and sftp push is already insane.
284
# FIXME: repository should inform if this is needed.
285
self.to_repository.reconcile()
288
class KnitRepoFetcher(RepoFetcher):
289
"""This is a knit format repository specific fetcher.
291
This differs from the GenericRepoFetcher by not doing a
292
reconciliation after copying, and using knit joining to
296
def _fetch_revision_texts(self, revs, pb):
297
# may need to be a InterRevisionStore call here.
298
from_transaction = self.from_repository.get_transaction()
299
to_transaction = self.to_repository.get_transaction()
300
to_sf = self.to_repository._revision_store.get_signature_file(
302
from_sf = self.from_repository._revision_store.get_signature_file(
304
to_sf.join(from_sf, version_ids=revs, ignore_missing=True)
305
to_rf = self.to_repository._revision_store.get_revision_file(
307
from_rf = self.from_repository._revision_store.get_revision_file(
309
to_rf.join(from_rf, version_ids=revs)
312
class Inter1and2Helper(object):
313
"""Helper for operations that convert data from model 1 and 2
315
This is for use by fetchers and converters.
318
def __init__(self, source, target):
321
:param source: The repository data comes from
322
:param target: The repository data goes to
327
def iter_rev_trees(self, revs):
328
"""Iterate through RevisionTrees efficiently.
330
Additionally, the inventory's revision_id is set if unset.
332
Trees are retrieved in batches of 100, and then yielded in the order
335
:param revs: A list of revision ids
337
# In case that revs is not a list.
340
for tree in self.source.revision_trees(revs[:100]):
341
if tree.inventory.revision_id is None:
342
tree.inventory.revision_id = tree.get_revision_id()
346
def generate_root_texts(self, revs):
347
"""Generate VersionedFiles for all root ids.
349
:param revs: the revisions to include
351
inventory_weave = self.source.get_inventory_weave()
354
to_store = self.target.weave_store
355
for tree in self.iter_rev_trees(revs):
356
revision_id = tree.inventory.root.revision
357
root_id = tree.get_root_id()
358
parents = inventory_weave.get_parents(revision_id)
359
if root_id not in versionedfile:
360
versionedfile[root_id] = to_store.get_weave_or_empty(root_id,
361
self.target.get_transaction())
362
_, _, parent_texts[root_id] = versionedfile[root_id].add_lines(
363
revision_id, parents, [], parent_texts)
365
def regenerate_inventory(self, revs):
366
"""Generate a new inventory versionedfile in target, convertin data.
368
The inventory is retrieved from the source, (deserializing it), and
369
stored in the target (reserializing it in a different format).
370
:param revs: The revisions to include
372
for tree in self.iter_rev_trees(revs):
373
parents = tree.get_parent_ids()
374
self.target.add_inventory(tree.get_revision_id(), tree.inventory,
378
class Model1toKnit2Fetcher(GenericRepoFetcher):
379
"""Fetch from a Model1 repository into a Knit2 repository
381
def __init__(self, to_repository, from_repository, last_revision=None,
382
pb=None, find_ghosts=True):
383
self.helper = Inter1and2Helper(from_repository, to_repository)
384
GenericRepoFetcher.__init__(self, to_repository, from_repository,
385
last_revision, pb, find_ghosts)
387
def _generate_root_texts(self, revs):
388
self.helper.generate_root_texts(revs)
390
def _fetch_inventory_weave(self, revs, pb):
391
self.helper.regenerate_inventory(revs)
394
class Knit1to2Fetcher(KnitRepoFetcher):
395
"""Fetch from a Knit1 repository into a Knit2 repository"""
397
def __init__(self, to_repository, from_repository, last_revision=None,
398
pb=None, find_ghosts=True):
399
self.helper = Inter1and2Helper(from_repository, to_repository)
400
KnitRepoFetcher.__init__(self, to_repository, from_repository,
401
last_revision, pb, find_ghosts)
403
def _generate_root_texts(self, revs):
404
self.helper.generate_root_texts(revs)
406
def _fetch_inventory_weave(self, revs, pb):
407
self.helper.regenerate_inventory(revs)
410
class RemoteToOtherFetcher(GenericRepoFetcher):
412
def _fetch_everything_for_revisions(self, revs, pp):
413
data_stream = self.from_repository.get_data_stream(revs)
414
self.to_repository.insert_data_stream(data_stream)
added
removed
bzrlib/fetch.py
