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
graph = self.from_repository.get_graph()
162
revs = list(graph.iter_topo_order(revs))
163
data_to_fetch = self.from_repository.item_keys_introduced_by(revs,
165
for knit_kind, file_id, revisions in data_to_fetch:
166
if knit_kind != phase:
168
# Make a new progress bar for this phase
171
pb = bzrlib.ui.ui_factory.nested_progress_bar()
172
if knit_kind == "file":
173
self._fetch_weave_text(file_id, revisions)
174
elif knit_kind == "inventory":
175
# Before we process the inventory we generate the root
176
# texts (if necessary) so that the inventories references
178
self._generate_root_texts(revs)
179
# NB: This currently reopens the inventory weave in source;
180
# using a full get_data_stream instead would avoid this.
181
self._fetch_inventory_weave(revs, pb)
182
elif knit_kind == "signatures":
183
# Nothing to do here; this will be taken care of when
184
# _fetch_revision_texts happens.
186
elif knit_kind == "revisions":
187
self._fetch_revision_texts(revs, pb)
189
raise AssertionError("Unknown knit kind %r" % knit_kind)
193
self.count_copied += len(revs)
195
def _revids_to_fetch(self):
196
"""Determines the exact revisions needed from self.from_repository to
197
install self._last_revision in self.to_repository.
199
If no revisions need to be fetched, then this just returns None.
201
mutter('fetch up to rev {%s}', self._last_revision)
202
if self._last_revision is NULL_REVISION:
203
# explicit limit of no revisions needed
205
if (self._last_revision is not None and
206
self.to_repository.has_revision(self._last_revision)):
209
return self.to_repository.search_missing_revision_ids(
210
self.from_repository, self._last_revision,
211
find_ghosts=self.find_ghosts)
212
except errors.NoSuchRevision:
213
raise InstallFailed([self._last_revision])
215
def _fetch_weave_text(self, file_id, required_versions):
216
to_weave = self.to_weaves.get_weave_or_empty(file_id,
217
self.to_repository.get_transaction())
218
from_weave = self.from_weaves.get_weave(file_id,
219
self.from_repository.get_transaction())
220
# we fetch all the texts, because texts do
221
# not reference anything, and its cheap enough
222
to_weave.join(from_weave, version_ids=required_versions)
224
def _fetch_inventory_weave(self, revs, pb):
225
pb.update("fetch inventory", 0, 2)
226
to_weave = self.to_repository.get_inventory_weave()
227
child_pb = bzrlib.ui.ui_factory.nested_progress_bar()
229
# just merge, this is optimisable and its means we don't
230
# copy unreferenced data such as not-needed inventories.
231
pb.update("fetch inventory", 1, 3)
232
from_weave = self.from_repository.get_inventory_weave()
233
pb.update("fetch inventory", 2, 3)
234
# we fetch only the referenced inventories because we do not
235
# know for unselected inventories whether all their required
236
# texts are present in the other repository - it could be
238
to_weave.join(from_weave, pb=child_pb, msg='merge inventory',
243
def _generate_root_texts(self, revs):
244
"""This will be called by __fetch between fetching weave texts and
245
fetching the inventory weave.
247
Subclasses should override this if they need to generate root texts
248
after fetching weave texts.
253
class GenericRepoFetcher(RepoFetcher):
254
"""This is a generic repo to repo fetcher.
256
This makes minimal assumptions about repo layout and contents.
257
It triggers a reconciliation after fetching to ensure integrity.
260
def _fetch_revision_texts(self, revs, pb):
261
"""Fetch revision object texts"""
262
to_txn = self.to_transaction = self.to_repository.get_transaction()
265
to_store = self.to_repository._revision_store
267
pb.update('copying revisions', count, total)
269
sig_text = self.from_repository.get_signature_text(rev)
270
to_store.add_revision_signature_text(rev, sig_text, to_txn)
271
except errors.NoSuchRevision:
274
to_store.add_revision(self.from_repository.get_revision(rev),
277
# fixup inventory if needed:
278
# this is expensive because we have no inverse index to current ghosts.
279
# but on local disk its a few seconds and sftp push is already insane.
281
# FIXME: repository should inform if this is needed.
282
self.to_repository.reconcile()
285
class KnitRepoFetcher(RepoFetcher):
286
"""This is a knit format repository specific fetcher.
288
This differs from the GenericRepoFetcher by not doing a
289
reconciliation after copying, and using knit joining to
293
def _fetch_revision_texts(self, revs, pb):
294
# may need to be a InterRevisionStore call here.
295
from_transaction = self.from_repository.get_transaction()
296
to_transaction = self.to_repository.get_transaction()
297
to_sf = self.to_repository._revision_store.get_signature_file(
299
from_sf = self.from_repository._revision_store.get_signature_file(
301
to_sf.join(from_sf, version_ids=revs, ignore_missing=True)
302
to_rf = self.to_repository._revision_store.get_revision_file(
304
from_rf = self.from_repository._revision_store.get_revision_file(
306
to_rf.join(from_rf, version_ids=revs)
309
class Inter1and2Helper(object):
310
"""Helper for operations that convert data from model 1 and 2
312
This is for use by fetchers and converters.
315
def __init__(self, source, target):
318
:param source: The repository data comes from
319
:param target: The repository data goes to
324
def iter_rev_trees(self, revs):
325
"""Iterate through RevisionTrees efficiently.
327
Additionally, the inventory's revision_id is set if unset.
329
Trees are retrieved in batches of 100, and then yielded in the order
332
:param revs: A list of revision ids
334
# In case that revs is not a list.
337
for tree in self.source.revision_trees(revs[:100]):
338
if tree.inventory.revision_id is None:
339
tree.inventory.revision_id = tree.get_revision_id()
343
def generate_root_texts(self, revs):
344
"""Generate VersionedFiles for all root ids.
346
:param revs: the revisions to include
348
inventory_weave = self.source.get_inventory_weave()
350
to_store = self.target.weave_store
351
graph = self.source.get_graph()
352
parent_map = graph.get_parent_map(revs)
353
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 revision 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
for root_id, versions in planned_versions.iteritems():
371
versionedfile = to_store.get_weave_or_empty(root_id,
372
self.target.get_transaction())
374
for revision_id in versions:
375
if revision_id in versionedfile:
377
parents = parent_map[revision_id]
378
# When a parent revision is a ghost, we guess that its root id
380
parents = tuple(p for p in parents if p != NULL_REVISION
381
and revision_root.get(p, root_id) == root_id)
382
result = versionedfile.add_lines_with_ghosts(
383
revision_id, parents, [], parent_texts)
384
parent_texts[revision_id] = result[2]
386
def regenerate_inventory(self, revs):
387
"""Generate a new inventory versionedfile in target, convertin data.
389
The inventory is retrieved from the source, (deserializing it), and
390
stored in the target (reserializing it in a different format).
391
:param revs: The revisions to include
393
for tree in self.iter_rev_trees(revs):
394
parents = tree.get_parent_ids()
395
self.target.add_inventory(tree.get_revision_id(), tree.inventory,
399
class Model1toKnit2Fetcher(GenericRepoFetcher):
400
"""Fetch from a Model1 repository into a Knit2 repository
402
def __init__(self, to_repository, from_repository, last_revision=None,
403
pb=None, find_ghosts=True):
404
self.helper = Inter1and2Helper(from_repository, to_repository)
405
GenericRepoFetcher.__init__(self, to_repository, from_repository,
406
last_revision, pb, find_ghosts)
408
def _generate_root_texts(self, revs):
409
self.helper.generate_root_texts(revs)
411
def _fetch_inventory_weave(self, revs, pb):
412
self.helper.regenerate_inventory(revs)
415
class Knit1to2Fetcher(KnitRepoFetcher):
416
"""Fetch from a Knit1 repository into a Knit2 repository"""
418
def __init__(self, to_repository, from_repository, last_revision=None,
419
pb=None, find_ghosts=True):
420
self.helper = Inter1and2Helper(from_repository, to_repository)
421
KnitRepoFetcher.__init__(self, to_repository, from_repository,
422
last_revision, pb, find_ghosts)
424
def _generate_root_texts(self, revs):
425
self.helper.generate_root_texts(revs)
427
def _fetch_inventory_weave(self, revs, pb):
428
self.helper.regenerate_inventory(revs)
431
class RemoteToOtherFetcher(GenericRepoFetcher):
433
def _fetch_everything_for_search(self, search, pp):
434
data_stream = self.from_repository.get_data_stream_for_search(search)
435
self.to_repository.insert_data_stream(data_stream)
added
removed
bzrlib/fetch.py
