1
# Copyright (C) 2005, 2006, 2008 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.
36
import bzrlib.errors as errors
37
from bzrlib.errors import InstallFailed
38
from bzrlib.progress import ProgressPhase
39
from bzrlib.revision import NULL_REVISION
40
from bzrlib.tsort import topo_sort
41
from bzrlib.trace import mutter
43
from bzrlib.versionedfile import FulltextContentFactory
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.
71
This should not be used directly, it's essential a object to encapsulate
72
the logic in InterRepository.fetch().
75
def __init__(self, to_repository, from_repository, last_revision=None, pb=None,
77
"""Create a repo fetcher.
79
:param find_ghosts: If True search the entire history for ghosts.
80
:param _write_group_acquired_callable: Don't use; this parameter only
81
exists to facilitate a hack done in InterPackRepo.fetch. We would
82
like to remove this parameter.
86
self.failed_revisions = []
87
if to_repository.has_same_location(from_repository):
88
# repository.fetch should be taking care of this case.
89
raise errors.BzrError('RepoFetcher run '
90
'between two objects at the same location: '
91
'%r and %r' % (to_repository, from_repository))
92
self.to_repository = to_repository
93
self.from_repository = from_repository
94
self.sink = to_repository._get_sink()
95
# must not mutate self._last_revision as its potentially a shared instance
96
self._last_revision = last_revision
97
self.find_ghosts = find_ghosts
99
self.pb = bzrlib.ui.ui_factory.nested_progress_bar()
100
self.nested_pb = self.pb
103
self.nested_pb = None
104
self.from_repository.lock_read()
109
if self.nested_pb is not None:
110
self.nested_pb.finished()
112
self.from_repository.unlock()
115
"""Primary worker function.
117
This initialises all the needed variables, and then fetches the
118
requested revisions, finally clearing the progress bar.
120
# Roughly this is what we're aiming for fetch to become:
122
# missing = self.sink.insert_stream(self.source.get_stream(search))
124
# missing = self.sink.insert_stream(self.source.get_items(missing))
127
self.file_ids_names = {}
128
pp = ProgressPhase('Transferring', 4, self.pb)
131
search = self._revids_to_fetch()
134
self._fetch_everything_for_search(search, pp)
138
def _fetch_everything_for_search(self, search, pp):
139
"""Fetch all data for the given set of revisions."""
140
# The first phase is "file". We pass the progress bar for it directly
141
# into item_keys_introduced_by, which has more information about how
142
# that phase is progressing than we do. Progress updates for the other
143
# phases are taken care of in this function.
144
# XXX: there should be a clear owner of the progress reporting. Perhaps
145
# item_keys_introduced_by should have a richer API than it does at the
146
# moment, so that it can feed the progress information back to this
148
if (self.from_repository._format.rich_root_data and
149
not self.to_repository._format.rich_root_data):
150
raise errors.IncompatibleRepositories(
151
self.from_repository, self.to_repository,
152
"different rich-root support")
153
self.pb = bzrlib.ui.ui_factory.nested_progress_bar()
155
source = self.from_repository._get_source(
156
self.to_repository._format)
157
stream = source.get_stream(search)
158
from_format = self.from_repository._format
159
resume_tokens, missing_keys = self.sink.insert_stream(
160
stream, from_format, [])
162
stream = source.get_stream_for_missing_keys(missing_keys)
163
resume_tokens, missing_keys = self.sink.insert_stream(
164
stream, from_format, resume_tokens)
166
raise AssertionError(
167
"second push failed to complete a fetch %r." % (
170
raise AssertionError(
171
"second push failed to commit the fetch %r." % (
174
self.count_copied = source.count_copied
176
if self.pb is not None:
179
def _revids_to_fetch(self):
180
"""Determines the exact revisions needed from self.from_repository to
181
install self._last_revision in self.to_repository.
183
If no revisions need to be fetched, then this just returns None.
185
mutter('fetch up to rev {%s}', self._last_revision)
186
if self._last_revision is NULL_REVISION:
187
# explicit limit of no revisions needed
189
if (self._last_revision is not None and
190
self.to_repository.has_revision(self._last_revision)):
193
return self.to_repository.search_missing_revision_ids(
194
self.from_repository, self._last_revision,
195
find_ghosts=self.find_ghosts)
196
except errors.NoSuchRevision, e:
197
raise InstallFailed([self._last_revision])
200
class Inter1and2Helper(object):
201
"""Helper for operations that convert data from model 1 and 2
203
This is for use by fetchers and converters.
206
def __init__(self, source):
209
:param source: The repository data comes from
213
def iter_rev_trees(self, revs):
214
"""Iterate through RevisionTrees efficiently.
216
Additionally, the inventory's revision_id is set if unset.
218
Trees are retrieved in batches of 100, and then yielded in the order
221
:param revs: A list of revision ids
223
# In case that revs is not a list.
226
for tree in self.source.revision_trees(revs[:100]):
227
if tree.inventory.revision_id is None:
228
tree.inventory.revision_id = tree.get_revision_id()
232
def _find_root_ids(self, revs, parent_map, graph):
234
planned_versions = {}
235
for tree in self.iter_rev_trees(revs):
236
revision_id = tree.inventory.root.revision
237
root_id = tree.get_root_id()
238
planned_versions.setdefault(root_id, []).append(revision_id)
239
revision_root[revision_id] = root_id
240
# Find out which parents we don't already know root ids for
242
for revision_parents in parent_map.itervalues():
243
parents.update(revision_parents)
244
parents.difference_update(revision_root.keys() + [NULL_REVISION])
245
# Limit to revisions present in the versionedfile
246
parents = graph.get_parent_map(parents).keys()
247
for tree in self.iter_rev_trees(parents):
248
root_id = tree.get_root_id()
249
revision_root[tree.get_revision_id()] = root_id
250
return revision_root, planned_versions
252
def generate_root_texts(self, revs):
253
"""Generate VersionedFiles for all root ids.
255
:param revs: the revisions to include
257
graph = self.source.get_graph()
258
parent_map = graph.get_parent_map(revs)
259
rev_order = topo_sort(parent_map)
260
rev_id_to_root_id, root_id_to_rev_ids = self._find_root_ids(
261
revs, parent_map, graph)
262
root_id_order = [(rev_id_to_root_id[rev_id], rev_id) for rev_id in
264
# Guaranteed stable, this groups all the file id operations together
265
# retaining topological order within the revisions of a file id.
266
# File id splits and joins would invalidate this, but they don't exist
267
# yet, and are unlikely to in non-rich-root environments anyway.
268
root_id_order.sort(key=operator.itemgetter(0))
269
# Create a record stream containing the roots to create.
271
for key in root_id_order:
272
root_id, rev_id = key
273
rev_parents = parent_map[rev_id]
274
# We drop revision parents with different file-ids, because
275
# that represents a rename of the root to a different location
276
# - its not actually a parent for us. (We could look for that
277
# file id in the revision tree at considerably more expense,
278
# but for now this is sufficient (and reconcile will catch and
279
# correct this anyway).
280
# When a parent revision is a ghost, we guess that its root id
281
# was unchanged (rather than trimming it from the parent list).
282
parent_keys = tuple((root_id, parent) for parent in rev_parents
283
if parent != NULL_REVISION and
284
rev_id_to_root_id.get(parent, root_id) == root_id)
285
yield FulltextContentFactory(key, parent_keys, None, '')
286
return [('texts', yield_roots())]
added
removed
bzrlib/fetch.py
