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()
115
if self.nested_pb is not None:
116
self.nested_pb.finished()
117
self.to_repository.unlock()
119
self.from_repository.unlock()
122
"""Primary worker function.
124
This initialises all the needed variables, and then fetches the
125
requested revisions, finally clearing the progress bar.
127
self.to_weaves = self.to_repository.weave_store
128
self.from_weaves = self.from_repository.weave_store
130
self.file_ids_names = {}
131
pp = ProgressPhase('Transferring', 4, self.pb)
134
search = self._revids_to_fetch()
137
if getattr(self, '_fetch_everything_for_search', None) is not None:
138
self._fetch_everything_for_search(search, pp)
140
# backward compatibility
141
self._fetch_everything_for_revisions(search.get_keys, pp)
145
def _fetch_everything_for_search(self, search, pp):
146
"""Fetch all data for the given set of revisions."""
147
# The first phase is "file". We pass the progress bar for it directly
148
# into item_keys_introduced_by, which has more information about how
149
# that phase is progressing than we do. Progress updates for the other
150
# phases are taken care of in this function.
151
# XXX: there should be a clear owner of the progress reporting. Perhaps
152
# item_keys_introduced_by should have a richer API than it does at the
153
# moment, so that it can feed the progress information back to this
156
pb = bzrlib.ui.ui_factory.nested_progress_bar()
158
revs = search.get_keys()
159
data_to_fetch = self.from_repository.item_keys_introduced_by(revs, pb)
160
for knit_kind, file_id, revisions in data_to_fetch:
161
if knit_kind != phase:
163
# Make a new progress bar for this phase
166
pb = bzrlib.ui.ui_factory.nested_progress_bar()
167
if knit_kind == "file":
168
self._fetch_weave_text(file_id, revisions)
169
elif knit_kind == "inventory":
170
# Before we process the inventory we generate the root
171
# texts (if necessary) so that the inventories references
173
self._generate_root_texts(revs)
174
# NB: This currently reopens the inventory weave in source;
175
# using a full get_data_stream instead would avoid this.
176
self._fetch_inventory_weave(revs, pb)
177
elif knit_kind == "signatures":
178
# Nothing to do here; this will be taken care of when
179
# _fetch_revision_texts happens.
181
elif knit_kind == "revisions":
182
self._fetch_revision_texts(revs, pb)
184
raise AssertionError("Unknown knit kind %r" % knit_kind)
188
self.count_copied += len(revs)
190
def _revids_to_fetch(self):
191
"""Determines the exact revisions needed from self.from_repository to
192
install self._last_revision in self.to_repository.
194
If no revisions need to be fetched, then this just returns None.
196
mutter('fetch up to rev {%s}', self._last_revision)
197
if self._last_revision is NULL_REVISION:
198
# explicit limit of no revisions needed
200
if (self._last_revision is not None and
201
self.to_repository.has_revision(self._last_revision)):
204
return self.to_repository.search_missing_revision_ids(
205
self.from_repository, self._last_revision,
206
find_ghosts=self.find_ghosts)
207
except errors.NoSuchRevision:
208
raise InstallFailed([self._last_revision])
210
def _fetch_weave_text(self, file_id, required_versions):
211
to_weave = self.to_weaves.get_weave_or_empty(file_id,
212
self.to_repository.get_transaction())
213
from_weave = self.from_weaves.get_weave(file_id,
214
self.from_repository.get_transaction())
215
# we fetch all the texts, because texts do
216
# not reference anything, and its cheap enough
217
to_weave.join(from_weave, version_ids=required_versions)
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',
238
def _generate_root_texts(self, revs):
239
"""This will be called by __fetch between fetching weave texts and
240
fetching the inventory weave.
242
Subclasses should override this if they need to generate root texts
243
after fetching weave texts.
248
class GenericRepoFetcher(RepoFetcher):
249
"""This is a generic repo to repo fetcher.
251
This makes minimal assumptions about repo layout and contents.
252
It triggers a reconciliation after fetching to ensure integrity.
255
def _fetch_revision_texts(self, revs, pb):
256
"""Fetch revision object texts"""
257
to_txn = self.to_transaction = self.to_repository.get_transaction()
260
to_store = self.to_repository._revision_store
262
pb.update('copying revisions', count, total)
264
sig_text = self.from_repository.get_signature_text(rev)
265
to_store.add_revision_signature_text(rev, sig_text, to_txn)
266
except errors.NoSuchRevision:
269
to_store.add_revision(self.from_repository.get_revision(rev),
272
# fixup inventory if needed:
273
# this is expensive because we have no inverse index to current ghosts.
274
# but on local disk its a few seconds and sftp push is already insane.
276
# FIXME: repository should inform if this is needed.
277
self.to_repository.reconcile()
280
class KnitRepoFetcher(RepoFetcher):
281
"""This is a knit format repository specific fetcher.
283
This differs from the GenericRepoFetcher by not doing a
284
reconciliation after copying, and using knit joining to
288
def _fetch_revision_texts(self, revs, pb):
289
# may need to be a InterRevisionStore call here.
290
from_transaction = self.from_repository.get_transaction()
291
to_transaction = self.to_repository.get_transaction()
292
to_sf = self.to_repository._revision_store.get_signature_file(
294
from_sf = self.from_repository._revision_store.get_signature_file(
296
to_sf.join(from_sf, version_ids=revs, ignore_missing=True)
297
to_rf = self.to_repository._revision_store.get_revision_file(
299
from_rf = self.from_repository._revision_store.get_revision_file(
301
to_rf.join(from_rf, version_ids=revs)
304
class Inter1and2Helper(object):
305
"""Helper for operations that convert data from model 1 and 2
307
This is for use by fetchers and converters.
310
def __init__(self, source, target):
313
:param source: The repository data comes from
314
:param target: The repository data goes to
319
def iter_rev_trees(self, revs):
320
"""Iterate through RevisionTrees efficiently.
322
Additionally, the inventory's revision_id is set if unset.
324
Trees are retrieved in batches of 100, and then yielded in the order
327
:param revs: A list of revision ids
329
# In case that revs is not a list.
332
for tree in self.source.revision_trees(revs[:100]):
333
if tree.inventory.revision_id is None:
334
tree.inventory.revision_id = tree.get_revision_id()
338
def generate_root_texts(self, revs):
339
"""Generate VersionedFiles for all root ids.
341
:param revs: the revisions to include
343
inventory_weave = self.source.get_inventory_weave()
346
to_store = self.target.weave_store
347
parent_map = self.source.get_graph().get_parent_map(revs)
348
for tree in self.iter_rev_trees(revs):
349
revision_id = tree.inventory.root.revision
350
root_id = tree.get_root_id()
351
parents = parent_map[revision_id]
352
if parents[0] == NULL_REVISION:
354
if root_id not in versionedfile:
355
versionedfile[root_id] = to_store.get_weave_or_empty(root_id,
356
self.target.get_transaction())
357
_, _, parent_texts[root_id] = versionedfile[root_id].add_lines(
358
revision_id, parents, [], parent_texts)
360
def regenerate_inventory(self, revs):
361
"""Generate a new inventory versionedfile in target, convertin data.
363
The inventory is retrieved from the source, (deserializing it), and
364
stored in the target (reserializing it in a different format).
365
:param revs: The revisions to include
367
for tree in self.iter_rev_trees(revs):
368
parents = tree.get_parent_ids()
369
self.target.add_inventory(tree.get_revision_id(), tree.inventory,
373
class Model1toKnit2Fetcher(GenericRepoFetcher):
374
"""Fetch from a Model1 repository into a Knit2 repository
376
def __init__(self, to_repository, from_repository, last_revision=None,
377
pb=None, find_ghosts=True):
378
self.helper = Inter1and2Helper(from_repository, to_repository)
379
GenericRepoFetcher.__init__(self, to_repository, from_repository,
380
last_revision, pb, find_ghosts)
382
def _generate_root_texts(self, revs):
383
self.helper.generate_root_texts(revs)
385
def _fetch_inventory_weave(self, revs, pb):
386
self.helper.regenerate_inventory(revs)
389
class Knit1to2Fetcher(KnitRepoFetcher):
390
"""Fetch from a Knit1 repository into a Knit2 repository"""
392
def __init__(self, to_repository, from_repository, last_revision=None,
393
pb=None, find_ghosts=True):
394
self.helper = Inter1and2Helper(from_repository, to_repository)
395
KnitRepoFetcher.__init__(self, to_repository, from_repository,
396
last_revision, pb, find_ghosts)
398
def _generate_root_texts(self, revs):
399
self.helper.generate_root_texts(revs)
401
def _fetch_inventory_weave(self, revs, pb):
402
self.helper.regenerate_inventory(revs)
405
class RemoteToOtherFetcher(GenericRepoFetcher):
407
def _fetch_everything_for_search(self, search, pp):
408
data_stream = self.from_repository.get_data_stream_for_search(search)
409
self.to_repository.insert_data_stream(data_stream)
added
removed
bzrlib/fetch.py
