13
13
# You should have received a copy of the GNU General Public License
14
14
# along with this program; if not, write to the Free Software
15
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, NoSuchRevision,
38
from bzrlib.trace import mutter
39
from bzrlib.progress import ProgressBar, ProgressPhase
40
from bzrlib.reconcile import RepoReconciler
41
from bzrlib.revision import NULL_REVISION
42
from bzrlib.symbol_versioning import *
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
@deprecated_function(zero_eight)
17
from bzrlib.trace import mutter, note
18
from bzrlib.branch import Branch
19
from bzrlib.progress import ProgressBar
23
def has_revision(branch, revision_id):
25
branch.get_revision_xml(revision_id)
27
except bzrlib.errors.NoSuchRevision:
64
31
def greedy_fetch(to_branch, from_branch, revision=None, pb=None):
65
"""Legacy API, please see branch.fetch(from_branch, last_revision, pb)."""
66
f = Fetcher(to_branch, from_branch, revision, pb)
67
return f.count_copied, f.failed_revisions
72
class RepoFetcher(object):
73
"""Pull revisions and texts from one repository to another.
76
if set, try to limit to the data this revision references.
79
count_copied -- number of revisions copied
81
This should not be used directory, its essential a object to encapsulate
82
the logic in InterRepository.fetch().
32
"""Copy a revision and all available ancestors from one branch to another
33
If no revision is specified, uses the last revision in the source branch's
84
def __init__(self, to_repository, from_repository, last_revision=None, pb=None):
86
self.failed_revisions = []
88
if to_repository.control_files._transport.base == from_repository.control_files._transport.base:
89
# check that last_revision is in 'from' and then return a no-operation.
90
if last_revision not in (None, NULL_REVISION):
91
from_repository.get_revision(last_revision)
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
98
self.pb = bzrlib.ui.ui_factory.nested_progress_bar()
99
self.nested_pb = self.pb
36
from_history = from_branch.revision_history()
37
required_revisions = set(from_history)
39
if revision is not None:
40
required_revisions.add(revision)
42
rev_index = from_history.index(revision)
45
if rev_index is not None:
46
from_history = from_history[:rev_index + 1]
102
self.nested_pb = None
103
self.from_repository.lock_read()
105
self.to_repository.lock_write()
109
if self.nested_pb is not None:
110
self.nested_pb.finished()
111
self.to_repository.unlock()
113
self.from_repository.unlock()
116
"""Primary worker function.
118
This initialises all the needed variables, and then fetches the
119
requested revisions, finally clearing the progress bar.
121
self.to_weaves = self.to_repository.weave_store
122
self.to_control = self.to_repository.control_weaves
123
self.from_weaves = self.from_repository.weave_store
124
self.from_control = self.from_repository.control_weaves
126
self.file_ids_names = {}
127
pp = ProgressPhase('fetch phase', 4, self.pb)
129
revs = self._revids_to_fetch()
133
self._fetch_weave_texts(revs)
135
self._fetch_inventory_weave(revs)
137
self._fetch_revision_texts(revs)
138
self.count_copied += len(revs)
142
def _revids_to_fetch(self):
143
mutter('fetch up to rev {%s}', self._last_revision)
144
if self._last_revision is NULL_REVISION:
145
# explicit limit of no revisions needed
147
if (self._last_revision != None and
148
self.to_repository.has_revision(self._last_revision)):
152
return self.to_repository.missing_revision_ids(self.from_repository,
154
except errors.NoSuchRevision:
155
raise InstallFailed([self._last_revision])
157
def _fetch_weave_texts(self, revs):
158
texts_pb = bzrlib.ui.ui_factory.nested_progress_bar()
160
file_ids = self.from_repository.fileids_altered_by_revision_ids(revs)
162
num_file_ids = len(file_ids)
163
for file_id, required_versions in file_ids.items():
164
texts_pb.update("fetch texts", count, num_file_ids)
166
to_weave = self.to_weaves.get_weave_or_empty(file_id,
167
self.to_repository.get_transaction())
168
from_weave = self.from_weaves.get_weave(file_id,
169
self.from_repository.get_transaction())
170
# we fetch all the texts, because texts do
171
# not reference anything, and its cheap enough
172
to_weave.join(from_weave, version_ids=required_versions)
173
# we don't need *all* of this data anymore, but we dont know
174
# what we do. This cache clearing will result in a new read
175
# of the knit data when we do the checkout, but probably we
176
# want to emit the needed data on the fly rather than at the
178
# the from weave should know not to cache data being joined,
179
# but its ok to ask it to clear.
180
from_weave.clear_cache()
181
to_weave.clear_cache()
185
def _fetch_inventory_weave(self, revs):
186
pb = bzrlib.ui.ui_factory.nested_progress_bar()
188
pb.update("fetch inventory", 0, 2)
189
to_weave = self.to_control.get_weave('inventory',
190
self.to_repository.get_transaction())
192
child_pb = bzrlib.ui.ui_factory.nested_progress_bar()
194
# just merge, this is optimisable and its means we don't
195
# copy unreferenced data such as not-needed inventories.
196
pb.update("fetch inventory", 1, 3)
197
from_weave = self.from_repository.get_inventory_weave()
198
pb.update("fetch inventory", 2, 3)
199
# we fetch only the referenced inventories because we do not
200
# know for unselected inventories whether all their required
201
# texts are present in the other repository - it could be
203
to_weave.join(from_weave, pb=child_pb, msg='merge inventory',
211
class GenericRepoFetcher(RepoFetcher):
212
"""This is a generic repo to repo fetcher.
214
This makes minimal assumptions about repo layout and contents.
215
It triggers a reconciliation after fetching to ensure integrity.
218
def _fetch_revision_texts(self, revs):
219
"""Fetch revision object texts"""
220
rev_pb = bzrlib.ui.ui_factory.nested_progress_bar()
222
to_txn = self.to_transaction = self.to_repository.get_transaction()
225
to_store = self.to_repository._revision_store
227
pb = bzrlib.ui.ui_factory.nested_progress_bar()
229
pb.update('copying revisions', count, total)
231
sig_text = self.from_repository.get_signature_text(rev)
232
to_store.add_revision_signature_text(rev, sig_text, to_txn)
233
except errors.NoSuchRevision:
236
to_store.add_revision(self.from_repository.get_revision(rev),
241
# fixup inventory if needed:
242
# this is expensive because we have no inverse index to current ghosts.
243
# but on local disk its a few seconds and sftp push is already insane.
245
# FIXME: repository should inform if this is needed.
246
self.to_repository.reconcile()
251
class KnitRepoFetcher(RepoFetcher):
252
"""This is a knit format repository specific fetcher.
254
This differs from the GenericRepoFetcher by not doing a
255
reconciliation after copying, and using knit joining to
259
def _fetch_revision_texts(self, revs):
260
# may need to be a InterRevisionStore call here.
261
from_transaction = self.from_repository.get_transaction()
262
to_transaction = self.to_repository.get_transaction()
263
to_sf = self.to_repository._revision_store.get_signature_file(
265
from_sf = self.from_repository._revision_store.get_signature_file(
267
to_sf.join(from_sf, version_ids=revs, ignore_missing=True)
268
to_rf = self.to_repository._revision_store.get_revision_file(
270
from_rf = self.from_repository._revision_store.get_revision_file(
272
to_rf.join(from_rf, version_ids=revs)
275
class Fetcher(object):
276
"""Backwards compatibility glue for branch.fetch()."""
278
@deprecated_method(zero_eight)
279
def __init__(self, to_branch, from_branch, last_revision=None, pb=None):
280
"""Please see branch.fetch()."""
281
to_branch.fetch(from_branch, last_revision, pb)
48
from_history = [revision]
49
to_history = to_branch.revision_history()
51
for rev_id in from_history:
52
if not has_revision(to_branch, rev_id):
53
missing.append(rev_id)
56
while len(missing) > 0:
57
installed, failed = to_branch.install_revisions(from_branch,
61
required_failed = failed.intersection(required_revisions)
62
if len(required_failed) > 0:
63
raise bzrlib.errors.InstallFailed(required_failed)
65
note("Failed to install %s" % rev_id)
66
all_failed.update(failed)
68
for rev_id in missing:
70
revision = from_branch.get_revision(rev_id)
71
except bzrlib.errors.NoSuchRevision:
72
if revision in from_history:
76
for parent in [p.revision_id for p in revision.parents]:
77
if not has_revision(to_branch, parent):
78
new_missing.add(parent)
80
return count, all_failed
added
removed
bzrlib/fetch.py
