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,
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 directory, its essential a object to encapsulate
76
the logic in InterRepository.fetch().
78
def __init__(self, to_repository, from_repository, last_revision=None, pb=None):
80
self.failed_revisions = []
82
if to_repository.has_same_location(from_repository):
83
# check that last_revision is in 'from' and then return a
85
if last_revision is not None and not is_null(last_revision):
86
to_repository.get_revision(last_revision)
88
self.to_repository = to_repository
89
self.from_repository = from_repository
90
# must not mutate self._last_revision as its potentially a shared instance
91
self._last_revision = last_revision
93
self.pb = bzrlib.ui.ui_factory.nested_progress_bar()
94
self.nested_pb = self.pb
98
self.from_repository.lock_read()
100
self.to_repository.lock_write()
102
self.to_repository.start_write_group()
106
self.to_repository.abort_write_group()
109
self.to_repository.commit_write_group()
111
if self.nested_pb is not None:
112
self.nested_pb.finished()
113
self.to_repository.unlock()
115
self.from_repository.unlock()
118
"""Primary worker function.
120
This initialises all the needed variables, and then fetches the
121
requested revisions, finally clearing the progress bar.
123
self.to_weaves = self.to_repository.weave_store
124
self.to_control = self.to_repository.control_weaves
125
self.from_weaves = self.from_repository.weave_store
126
self.from_control = self.from_repository.control_weaves
128
self.file_ids_names = {}
129
pp = ProgressPhase('Fetch phase', 4, self.pb)
132
revs = self._revids_to_fetch()
133
self._fetch_everything_for_revisions(revs, pp)
137
def _fetch_everything_for_revisions(self, revs, pp):
138
"""Fetch all data for the given set of revisions."""
141
# The first phase is "file". We pass the progress bar for it directly
142
# into item_keys_introduced_by, which has more information about how
143
# that phase is progressing than we do. Progress updates for the other
144
# phases are taken care of in this function.
145
# XXX: there should be a clear owner of the progress reporting. Perhaps
146
# item_keys_introduced_by should have a richer API than it does at the
147
# moment, so that it can feed the progress information back to this
150
pb = bzrlib.ui.ui_factory.nested_progress_bar()
152
data_to_fetch = self.from_repository.item_keys_introduced_by(revs, pb)
153
for knit_kind, file_id, revisions in data_to_fetch:
154
if knit_kind != phase:
156
# Make a new progress bar for this phase
159
pb = bzrlib.ui.ui_factory.nested_progress_bar()
160
if knit_kind == "file":
161
self._fetch_weave_text(file_id, revisions)
162
elif knit_kind == "inventory":
164
# Once we've processed all the files, then we generate the root
165
# texts (if necessary), then we process the inventory. It's a
166
# bit distasteful to have knit_kind == "inventory" mean this,
167
# perhaps it should happen on the first non-"file" knit, in case
168
# it's not always inventory?
169
self._generate_root_texts(revs)
170
self._fetch_inventory_weave(revs, pb)
171
elif knit_kind == "signatures":
172
# Nothing to do here; this will be taken care of when
173
# _fetch_revision_texts happens.
175
elif knit_kind == "revisions":
176
self._fetch_revision_texts(revs, pb)
178
raise AssertionError("Unknown knit kind %r" % knit_kind)
182
self.count_copied += len(revs)
184
def _revids_to_fetch(self):
185
"""Determines the exact revisions needed from self.from_repository to
186
install self._last_revision in self.to_repository.
188
If no revisions need to be fetched, then this just returns None.
190
mutter('fetch up to rev {%s}', self._last_revision)
191
if self._last_revision is NULL_REVISION:
192
# explicit limit of no revisions needed
194
if (self._last_revision is not None and
195
self.to_repository.has_revision(self._last_revision)):
199
return self.to_repository.missing_revision_ids(self.from_repository,
201
except errors.NoSuchRevision:
202
raise InstallFailed([self._last_revision])
204
def _fetch_weave_text(self, file_id, required_versions):
205
to_weave = self.to_weaves.get_weave_or_empty(file_id,
206
self.to_repository.get_transaction())
207
from_weave = self.from_weaves.get_weave(file_id,
208
self.from_repository.get_transaction())
209
# we fetch all the texts, because texts do
210
# not reference anything, and its cheap enough
211
to_weave.join(from_weave, version_ids=required_versions)
212
# we don't need *all* of this data anymore, but we dont know
213
# what we do. This cache clearing will result in a new read
214
# of the knit data when we do the checkout, but probably we
215
# want to emit the needed data on the fly rather than at the
217
# the from weave should know not to cache data being joined,
218
# but its ok to ask it to clear.
219
from_weave.clear_cache()
220
to_weave.clear_cache()
222
def _fetch_inventory_weave(self, revs, pb):
223
pb.update("fetch inventory", 0, 2)
224
to_weave = self.to_control.get_weave('inventory',
225
self.to_repository.get_transaction())
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',
240
from_weave.clear_cache()
244
def _generate_root_texts(self, revs):
245
"""This will be called by __fetch between fetching weave texts and
246
fetching the inventory weave.
248
Subclasses should override this if they need to generate root texts
249
after fetching weave texts.
254
class GenericRepoFetcher(RepoFetcher):
255
"""This is a generic repo to repo fetcher.
257
This makes minimal assumptions about repo layout and contents.
258
It triggers a reconciliation after fetching to ensure integrity.
261
def _fetch_revision_texts(self, revs, pb):
262
"""Fetch revision object texts"""
263
to_txn = self.to_transaction = self.to_repository.get_transaction()
266
to_store = self.to_repository._revision_store
268
pb.update('copying revisions', count, total)
270
sig_text = self.from_repository.get_signature_text(rev)
271
to_store.add_revision_signature_text(rev, sig_text, to_txn)
272
except errors.NoSuchRevision:
275
to_store.add_revision(self.from_repository.get_revision(rev),
278
# fixup inventory if needed:
279
# this is expensive because we have no inverse index to current ghosts.
280
# but on local disk its a few seconds and sftp push is already insane.
282
# FIXME: repository should inform if this is needed.
283
self.to_repository.reconcile()
286
class KnitRepoFetcher(RepoFetcher):
287
"""This is a knit format repository specific fetcher.
289
This differs from the GenericRepoFetcher by not doing a
290
reconciliation after copying, and using knit joining to
294
def _fetch_revision_texts(self, revs, pb):
295
# may need to be a InterRevisionStore call here.
296
from_transaction = self.from_repository.get_transaction()
297
to_transaction = self.to_repository.get_transaction()
298
to_sf = self.to_repository._revision_store.get_signature_file(
300
from_sf = self.from_repository._revision_store.get_signature_file(
302
to_sf.join(from_sf, version_ids=revs, ignore_missing=True)
303
to_rf = self.to_repository._revision_store.get_revision_file(
305
from_rf = self.from_repository._revision_store.get_revision_file(
307
to_rf.join(from_rf, version_ids=revs)
310
class Inter1and2Helper(object):
311
"""Helper for operations that convert data from model 1 and 2
313
This is for use by fetchers and converters.
316
def __init__(self, source, target):
319
:param source: The repository data comes from
320
:param target: The repository data goes to
325
def iter_rev_trees(self, revs):
326
"""Iterate through RevisionTrees efficiently.
328
Additionally, the inventory's revision_id is set if unset.
330
Trees are retrieved in batches of 100, and then yielded in the order
333
:param revs: A list of revision ids
336
for tree in self.source.revision_trees(revs[:100]):
337
if tree.inventory.revision_id is None:
338
tree.inventory.revision_id = tree.get_revision_id()
342
def generate_root_texts(self, revs):
343
"""Generate VersionedFiles for all root ids.
345
:param revs: the revisions to include
347
inventory_weave = self.source.get_inventory_weave()
350
to_store = self.target.weave_store
351
for tree in self.iter_rev_trees(revs):
352
revision_id = tree.inventory.root.revision
353
root_id = tree.inventory.root.file_id
354
parents = inventory_weave.get_parents(revision_id)
355
if root_id not in versionedfile:
356
versionedfile[root_id] = to_store.get_weave_or_empty(root_id,
357
self.target.get_transaction())
358
parent_texts[root_id] = versionedfile[root_id].add_lines(
359
revision_id, parents, [], parent_texts)
361
def regenerate_inventory(self, revs):
362
"""Generate a new inventory versionedfile in target, convertin data.
364
The inventory is retrieved from the source, (deserializing it), and
365
stored in the target (reserializing it in a different format).
366
:param revs: The revisions to include
368
inventory_weave = self.source.get_inventory_weave()
369
for tree in self.iter_rev_trees(revs):
370
parents = inventory_weave.get_parents(tree.get_revision_id())
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,
380
self.helper = Inter1and2Helper(from_repository, to_repository)
381
GenericRepoFetcher.__init__(self, to_repository, from_repository,
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,
396
self.helper = Inter1and2Helper(from_repository, to_repository)
397
KnitRepoFetcher.__init__(self, to_repository, from_repository,
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)
added
removed
bzrlib/fetch.py
