1
# Copyright (C) 2005 by Canonical Ltd
1
# Copyright (C) 2005, 2006, 2008 Canonical Ltd
3
3
# This program is free software; you can redistribute it and/or modify
4
4
# it under the terms of the GNU General Public License as published by
5
5
# the Free Software Foundation; either version 2 of the License, or
6
6
# (at your option) any later version.
8
8
# This program is distributed in the hope that it will be useful,
9
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
11
# GNU General Public License for more details.
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
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:
31
def greedy_fetch(to_branch, from_branch, revision=None, pb=None):
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
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.
69
This should not be used directly, it's essential a object to encapsulate
70
the logic in InterRepository.fetch().
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]
73
def __init__(self, to_repository, from_repository, last_revision=None, pb=None,
75
"""Create a repo fetcher.
77
:param find_ghosts: If True search the entire history for ghosts.
78
:param _write_group_acquired_callable: Don't use; this parameter only
79
exists to facilitate a hack done in InterPackRepo.fetch. We would
80
like to remove this parameter.
82
if to_repository.has_same_location(from_repository):
83
# repository.fetch should be taking care of this case.
84
raise errors.BzrError('RepoFetcher run '
85
'between two objects at the same location: '
86
'%r and %r' % (to_repository, from_repository))
87
self.to_repository = to_repository
88
self.from_repository = from_repository
89
self.sink = to_repository._get_sink()
90
# must not mutate self._last_revision as its potentially a shared instance
91
self._last_revision = last_revision
92
self.find_ghosts = find_ghosts
94
self.pb = bzrlib.ui.ui_factory.nested_progress_bar()
95
self.nested_pb = self.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:
99
self.from_repository.lock_read()
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
104
if self.nested_pb is not None:
105
self.nested_pb.finished()
107
self.from_repository.unlock()
110
"""Primary worker function.
112
This initialises all the needed variables, and then fetches the
113
requested revisions, finally clearing the progress bar.
115
# Roughly this is what we're aiming for fetch to become:
117
# missing = self.sink.insert_stream(self.source.get_stream(search))
119
# missing = self.sink.insert_stream(self.source.get_items(missing))
122
self.file_ids_names = {}
123
pp = ProgressPhase('Transferring', 4, self.pb)
126
search = self._revids_to_fetch()
129
self._fetch_everything_for_search(search, pp)
133
def _fetch_everything_for_search(self, search, pp):
134
"""Fetch all data for the given set of revisions."""
135
# The first phase is "file". We pass the progress bar for it directly
136
# into item_keys_introduced_by, which has more information about how
137
# that phase is progressing than we do. Progress updates for the other
138
# phases are taken care of in this function.
139
# XXX: there should be a clear owner of the progress reporting. Perhaps
140
# item_keys_introduced_by should have a richer API than it does at the
141
# moment, so that it can feed the progress information back to this
143
if (self.from_repository._format.rich_root_data and
144
not self.to_repository._format.rich_root_data):
145
raise errors.IncompatibleRepositories(
146
self.from_repository, self.to_repository,
147
"different rich-root support")
148
self.pb = bzrlib.ui.ui_factory.nested_progress_bar()
150
source = self.from_repository._get_source(
151
self.to_repository._format)
152
stream = source.get_stream(search)
153
from_format = self.from_repository._format
154
resume_tokens, missing_keys = self.sink.insert_stream(
155
stream, from_format, [])
157
stream = source.get_stream_for_missing_keys(missing_keys)
158
resume_tokens, missing_keys = self.sink.insert_stream(
159
stream, from_format, resume_tokens)
161
raise AssertionError(
162
"second push failed to complete a fetch %r." % (
165
raise AssertionError(
166
"second push failed to commit the fetch %r." % (
170
if self.pb is not None:
173
def _revids_to_fetch(self):
174
"""Determines the exact revisions needed from self.from_repository to
175
install self._last_revision in self.to_repository.
177
If no revisions need to be fetched, then this just returns None.
179
mutter('fetch up to rev {%s}', self._last_revision)
180
if self._last_revision is NULL_REVISION:
181
# explicit limit of no revisions needed
183
if (self._last_revision is not None and
184
self.to_repository.has_revision(self._last_revision)):
187
return self.to_repository.search_missing_revision_ids(
188
self.from_repository, self._last_revision,
189
find_ghosts=self.find_ghosts)
190
except errors.NoSuchRevision, e:
191
raise InstallFailed([self._last_revision])
194
class Inter1and2Helper(object):
195
"""Helper for operations that convert data from model 1 and 2
197
This is for use by fetchers and converters.
200
def __init__(self, source):
203
:param source: The repository data comes from
207
def iter_rev_trees(self, revs):
208
"""Iterate through RevisionTrees efficiently.
210
Additionally, the inventory's revision_id is set if unset.
212
Trees are retrieved in batches of 100, and then yielded in the order
215
:param revs: A list of revision ids
217
# In case that revs is not a list.
220
for tree in self.source.revision_trees(revs[:100]):
221
if tree.inventory.revision_id is None:
222
tree.inventory.revision_id = tree.get_revision_id()
226
def _find_root_ids(self, revs, parent_map, graph):
228
planned_versions = {}
229
for tree in self.iter_rev_trees(revs):
230
revision_id = tree.inventory.root.revision
231
root_id = tree.get_root_id()
232
planned_versions.setdefault(root_id, []).append(revision_id)
233
revision_root[revision_id] = root_id
234
# Find out which parents we don't already know root ids for
236
for revision_parents in parent_map.itervalues():
237
parents.update(revision_parents)
238
parents.difference_update(revision_root.keys() + [NULL_REVISION])
239
# Limit to revisions present in the versionedfile
240
parents = graph.get_parent_map(parents).keys()
241
for tree in self.iter_rev_trees(parents):
242
root_id = tree.get_root_id()
243
revision_root[tree.get_revision_id()] = root_id
244
return revision_root, planned_versions
246
def generate_root_texts(self, revs):
247
"""Generate VersionedFiles for all root ids.
249
:param revs: the revisions to include
251
graph = self.source.get_graph()
252
parent_map = graph.get_parent_map(revs)
253
rev_order = topo_sort(parent_map)
254
rev_id_to_root_id, root_id_to_rev_ids = self._find_root_ids(
255
revs, parent_map, graph)
256
root_id_order = [(rev_id_to_root_id[rev_id], rev_id) for rev_id in
258
# Guaranteed stable, this groups all the file id operations together
259
# retaining topological order within the revisions of a file id.
260
# File id splits and joins would invalidate this, but they don't exist
261
# yet, and are unlikely to in non-rich-root environments anyway.
262
root_id_order.sort(key=operator.itemgetter(0))
263
# Create a record stream containing the roots to create.
265
for key in root_id_order:
266
root_id, rev_id = key
267
rev_parents = parent_map[rev_id]
268
# We drop revision parents with different file-ids, because
269
# that represents a rename of the root to a different location
270
# - its not actually a parent for us. (We could look for that
271
# file id in the revision tree at considerably more expense,
272
# but for now this is sufficient (and reconcile will catch and
273
# correct this anyway).
274
# When a parent revision is a ghost, we guess that its root id
275
# was unchanged (rather than trimming it from the parent list).
276
parent_keys = tuple((root_id, parent) for parent in rev_parents
277
if parent != NULL_REVISION and
278
rev_id_to_root_id.get(parent, root_id) == root_id)
279
yield FulltextContentFactory(key, parent_keys, None, '')
280
return [('texts', yield_roots())]
added
removed
bzrlib/fetch.py
