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
43
from bzrlib.versionedfile import filter_absent
45
# TODO: Avoid repeatedly opening weaves so many times.
47
# XXX: This doesn't handle ghost (not present in branch) revisions at
48
# all yet. I'm not sure they really should be supported.
50
# NOTE: This doesn't copy revisions which may be present but not
51
# merged into the last revision. I'm not sure we want to do that.
53
# - get a list of revisions that need to be pulled in
54
# - for each one, pull in that revision file
55
# and get the inventory, and store the inventory with right
57
# - and get the ancestry, and store that with right parents too
58
# - and keep a note of all file ids and version seen
59
# - then go through all files; for each one get the weave,
60
# and add in all file versions
63
class RepoFetcher(object):
64
"""Pull revisions and texts from one repository to another.
67
if set, try to limit to the data this revision references.
70
count_copied -- number of revisions copied
72
This should not be used directly, it's essential a object to encapsulate
73
the logic in InterRepository.fetch().
76
def __init__(self, to_repository, from_repository, last_revision=None, pb=None,
78
"""Create a repo fetcher.
80
:param find_ghosts: If True search the entire history for ghosts.
83
self.failed_revisions = []
85
if to_repository.has_same_location(from_repository):
86
# repository.fetch should be taking care of this case.
87
raise errors.BzrError('RepoFetcher run '
88
'between two objects at the same location: '
89
'%r and %r' % (to_repository, from_repository))
90
self.to_repository = to_repository
91
self.from_repository = from_repository
92
# must not mutate self._last_revision as its potentially a shared instance
93
self._last_revision = last_revision
94
self.find_ghosts = find_ghosts
96
self.pb = bzrlib.ui.ui_factory.nested_progress_bar()
97
self.nested_pb = self.pb
100
self.nested_pb = None
101
self.from_repository.lock_read()
103
self.to_repository.lock_write()
105
self.to_repository.start_write_group()
109
self.to_repository.abort_write_group()
112
self.to_repository.commit_write_group()
115
if self.nested_pb is not None:
116
self.nested_pb.finished()
118
self.to_repository.unlock()
120
self.from_repository.unlock()
123
"""Primary worker function.
125
This initialises all the needed variables, and then fetches the
126
requested revisions, finally clearing the progress bar.
128
self.to_weaves = self.to_repository.weave_store
129
self.from_weaves = self.from_repository.weave_store
131
self.file_ids_names = {}
132
pp = ProgressPhase('Transferring', 4, self.pb)
135
search = self._revids_to_fetch()
138
if getattr(self, '_fetch_everything_for_search', None) is not None:
139
self._fetch_everything_for_search(search, pp)
141
# backward compatibility
142
self._fetch_everything_for_revisions(search.get_keys, pp)
146
def _fetch_everything_for_search(self, search, pp):
147
"""Fetch all data for the given set of revisions."""
148
# The first phase is "file". We pass the progress bar for it directly
149
# into item_keys_introduced_by, which has more information about how
150
# that phase is progressing than we do. Progress updates for the other
151
# phases are taken care of in this function.
152
# XXX: there should be a clear owner of the progress reporting. Perhaps
153
# item_keys_introduced_by should have a richer API than it does at the
154
# moment, so that it can feed the progress information back to this
157
pb = bzrlib.ui.ui_factory.nested_progress_bar()
159
revs = search.get_keys()
160
graph = self.from_repository.get_graph()
161
revs = list(graph.iter_topo_order(revs))
162
data_to_fetch = self.from_repository.item_keys_introduced_by(revs,
164
for knit_kind, file_id, revisions in data_to_fetch:
165
if knit_kind != phase:
167
# Make a new progress bar for this phase
170
pb = bzrlib.ui.ui_factory.nested_progress_bar()
171
if knit_kind == "file":
172
self._fetch_weave_text(file_id, revisions)
173
elif knit_kind == "inventory":
174
# Before we process the inventory we generate the root
175
# texts (if necessary) so that the inventories references
177
self._generate_root_texts(revs)
178
# NB: This currently reopens the inventory weave in source;
179
# using a full get_data_stream instead would avoid this.
180
self._fetch_inventory_weave(revs, pb)
181
elif knit_kind == "signatures":
182
# Nothing to do here; this will be taken care of when
183
# _fetch_revision_texts happens.
185
elif knit_kind == "revisions":
186
self._fetch_revision_texts(revs, pb)
188
raise AssertionError("Unknown knit kind %r" % knit_kind)
192
self.count_copied += len(revs)
194
def _revids_to_fetch(self):
195
"""Determines the exact revisions needed from self.from_repository to
196
install self._last_revision in self.to_repository.
198
If no revisions need to be fetched, then this just returns None.
200
mutter('fetch up to rev {%s}', self._last_revision)
201
if self._last_revision is NULL_REVISION:
202
# explicit limit of no revisions needed
204
if (self._last_revision is not None and
205
self.to_repository.has_revision(self._last_revision)):
208
return self.to_repository.search_missing_revision_ids(
209
self.from_repository, self._last_revision,
210
find_ghosts=self.find_ghosts)
211
except errors.NoSuchRevision:
212
raise InstallFailed([self._last_revision])
214
def _fetch_weave_text(self, file_id, required_versions):
215
to_weave = self.to_weaves.get_weave_or_empty(file_id,
216
self.to_repository.get_transaction())
217
from_weave = self.from_weaves.get_weave(file_id,
218
self.from_repository.get_transaction())
219
# Fetch all the texts.
220
to_weave.insert_record_stream(from_weave.get_record_stream(
221
required_versions, 'topological', False))
223
def _fetch_inventory_weave(self, revs, pb):
224
pb.update("fetch inventory", 0, 2)
225
to_weave = self.to_repository.get_inventory_weave()
226
child_pb = bzrlib.ui.ui_factory.nested_progress_bar()
228
# just merge, this is optimisable and its means we don't
229
# copy unreferenced data such as not-needed inventories.
230
pb.update("fetch inventory", 1, 3)
231
from_weave = self.from_repository.get_inventory_weave()
232
pb.update("fetch inventory", 2, 3)
233
# we fetch only the referenced inventories because we do not
234
# know for unselected inventories whether all their required
235
# texts are present in the other repository - it could be
237
to_weave.insert_record_stream(from_weave.get_record_stream(revs,
238
'topological', False))
242
def _generate_root_texts(self, revs):
243
"""This will be called by __fetch between fetching weave texts and
244
fetching the inventory weave.
246
Subclasses should override this if they need to generate root texts
247
after fetching weave texts.
252
class GenericRepoFetcher(RepoFetcher):
253
"""This is a generic repo to repo fetcher.
255
This makes minimal assumptions about repo layout and contents.
256
It triggers a reconciliation after fetching to ensure integrity.
259
def _fetch_revision_texts(self, revs, pb):
260
"""Fetch revision object texts"""
261
to_txn = self.to_transaction = self.to_repository.get_transaction()
264
to_store = self.to_repository._revision_store
266
pb.update('copying revisions', count, total)
268
sig_text = self.from_repository.get_signature_text(rev)
269
to_store.add_revision_signature_text(rev, sig_text, to_txn)
270
except errors.NoSuchRevision:
273
self._copy_revision(rev, to_txn)
275
# fixup inventory if needed:
276
# this is expensive because we have no inverse index to current ghosts.
277
# but on local disk its a few seconds and sftp push is already insane.
279
# FIXME: repository should inform if this is needed.
280
self.to_repository.reconcile()
282
def _copy_revision(self, rev, to_txn):
283
to_store = self.to_repository._revision_store
284
to_store.add_revision(self.from_repository.get_revision(rev), to_txn)
287
class KnitRepoFetcher(RepoFetcher):
288
"""This is a knit format repository specific fetcher.
290
This differs from the GenericRepoFetcher by not doing a
291
reconciliation after copying, and using knit joining to
295
def _fetch_revision_texts(self, revs, pb):
296
# may need to be a InterRevisionStore call here.
297
from_transaction = self.from_repository.get_transaction()
298
to_transaction = self.to_repository.get_transaction()
299
to_sf = self.to_repository._revision_store.get_signature_file(
301
from_sf = self.from_repository._revision_store.get_signature_file(
303
# A missing signature is just skipped.
304
to_sf.insert_record_stream(filter_absent(from_sf.get_record_stream(revs,
305
'unordered', False)))
306
self._fetch_just_revision_texts(revs, from_transaction, to_transaction)
308
def _fetch_just_revision_texts(self, version_ids, from_transaction,
310
to_rf = self.to_repository._revision_store.get_revision_file(
312
from_rf = self.from_repository._revision_store.get_revision_file(
314
to_rf.insert_record_stream(from_rf.get_record_stream(version_ids,
315
'topological', False))
318
class Inter1and2Helper(object):
319
"""Helper for operations that convert data from model 1 and 2
321
This is for use by fetchers and converters.
324
def __init__(self, source, target):
327
:param source: The repository data comes from
328
:param target: The repository data goes to
333
def iter_rev_trees(self, revs):
334
"""Iterate through RevisionTrees efficiently.
336
Additionally, the inventory's revision_id is set if unset.
338
Trees are retrieved in batches of 100, and then yielded in the order
341
:param revs: A list of revision ids
343
# In case that revs is not a list.
346
for tree in self.source.revision_trees(revs[:100]):
347
if tree.inventory.revision_id is None:
348
tree.inventory.revision_id = tree.get_revision_id()
352
def _find_root_ids(self, revs, parent_map, graph):
354
planned_versions = {}
355
for tree in self.iter_rev_trees(revs):
356
revision_id = tree.inventory.root.revision
357
root_id = tree.get_root_id()
358
planned_versions.setdefault(root_id, []).append(revision_id)
359
revision_root[revision_id] = root_id
360
# Find out which parents we don't already know root ids for
362
for revision_parents in parent_map.itervalues():
363
parents.update(revision_parents)
364
parents.difference_update(revision_root.keys() + [NULL_REVISION])
365
# Limit to revisions present in the versionedfile
366
parents = graph.get_parent_map(parents).keys()
367
for tree in self.iter_rev_trees(parents):
368
root_id = tree.get_root_id()
369
revision_root[tree.get_revision_id()] = root_id
370
return revision_root, planned_versions
372
def generate_root_texts(self, revs):
373
"""Generate VersionedFiles for all root ids.
375
:param revs: the revisions to include
377
to_store = self.target.weave_store
378
graph = self.source.get_graph()
379
parent_map = graph.get_parent_map(revs)
380
revision_root, planned_versions = self._find_root_ids(
381
revs, parent_map, graph)
382
for root_id, versions in planned_versions.iteritems():
383
versionedfile = to_store.get_weave_or_empty(root_id,
384
self.target.get_transaction())
386
for revision_id in versions:
387
if revision_id in versionedfile:
389
parents = parent_map[revision_id]
390
# We drop revision parents with different file-ids, because
391
# a version cannot have a version with another file-id as its
393
# When a parent revision is a ghost, we guess that its root id
395
parents = tuple(p for p in parents if p != NULL_REVISION
396
and revision_root.get(p, root_id) == root_id)
397
result = versionedfile.add_lines_with_ghosts(
398
revision_id, parents, [], parent_texts)
399
parent_texts[revision_id] = result[2]
401
def regenerate_inventory(self, revs):
402
"""Generate a new inventory versionedfile in target, convertin data.
404
The inventory is retrieved from the source, (deserializing it), and
405
stored in the target (reserializing it in a different format).
406
:param revs: The revisions to include
408
for tree in self.iter_rev_trees(revs):
409
parents = tree.get_parent_ids()
410
self.target.add_inventory(tree.get_revision_id(), tree.inventory,
413
def fetch_revisions(self, revision_ids):
414
for revision in self.source.get_revisions(revision_ids):
415
self.target.add_revision(revision.revision_id, revision)
418
class Model1toKnit2Fetcher(GenericRepoFetcher):
419
"""Fetch from a Model1 repository into a Knit2 repository
421
def __init__(self, to_repository, from_repository, last_revision=None,
422
pb=None, find_ghosts=True):
423
self.helper = Inter1and2Helper(from_repository, to_repository)
424
GenericRepoFetcher.__init__(self, to_repository, from_repository,
425
last_revision, pb, find_ghosts)
427
def _generate_root_texts(self, revs):
428
self.helper.generate_root_texts(revs)
430
def _fetch_inventory_weave(self, revs, pb):
431
self.helper.regenerate_inventory(revs)
433
def _copy_revision(self, rev, to_txn):
434
self.helper.fetch_revisions([rev])
437
class Knit1to2Fetcher(KnitRepoFetcher):
438
"""Fetch from a Knit1 repository into a Knit2 repository"""
440
def __init__(self, to_repository, from_repository, last_revision=None,
441
pb=None, find_ghosts=True):
442
self.helper = Inter1and2Helper(from_repository, to_repository)
443
KnitRepoFetcher.__init__(self, to_repository, from_repository,
444
last_revision, pb, find_ghosts)
446
def _generate_root_texts(self, revs):
447
self.helper.generate_root_texts(revs)
449
def _fetch_inventory_weave(self, revs, pb):
450
self.helper.regenerate_inventory(revs)
452
def _fetch_just_revision_texts(self, version_ids, from_transaction,
454
self.helper.fetch_revisions(version_ids)
457
class RemoteToOtherFetcher(GenericRepoFetcher):
459
def _fetch_everything_for_search(self, search, pp):
460
data_stream = self.from_repository.get_data_stream_for_search(search)
461
self.count_copied += self.to_repository.insert_data_stream(data_stream)
added
removed
bzrlib/fetch.py
