1
# Copyright (C) 2005, 2006, 2008, 2009 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., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 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
28
from bzrlib.lazy_import import lazy_import
29
lazy_import(globals(), """
43
from bzrlib.revision import NULL_REVISION
44
from bzrlib.trace import mutter
47
class RepoFetcher(object):
48
"""Pull revisions and texts from one repository to another.
50
This should not be used directly, it's essential a object to encapsulate
51
the logic in InterRepository.fetch().
54
def __init__(self, to_repository, from_repository, last_revision=None,
55
find_ghosts=True, fetch_spec=None):
56
"""Create a repo fetcher.
58
:param last_revision: If set, try to limit to the data this revision
60
:param find_ghosts: If True search the entire history for ghosts.
62
# repository.fetch has the responsibility for short-circuiting
63
# attempts to copy between a repository and itself.
64
self.to_repository = to_repository
65
self.from_repository = from_repository
66
self.sink = to_repository._get_sink()
67
# must not mutate self._last_revision as its potentially a shared instance
68
self._last_revision = last_revision
69
self._fetch_spec = fetch_spec
70
self.find_ghosts = find_ghosts
71
self.from_repository.lock_read()
72
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
73
self.from_repository, self.from_repository._format,
74
self.to_repository, self.to_repository._format)
78
self.from_repository.unlock()
81
"""Primary worker function.
83
This initialises all the needed variables, and then fetches the
84
requested revisions, finally clearing the progress bar.
86
# Roughly this is what we're aiming for fetch to become:
88
# missing = self.sink.insert_stream(self.source.get_stream(search))
90
# missing = self.sink.insert_stream(self.source.get_items(missing))
93
self.file_ids_names = {}
94
pb = ui.ui_factory.nested_progress_bar()
95
pb.show_pct = pb.show_count = False
97
pb.update("Finding revisions", 0, 2)
98
search = self._revids_to_fetch()
101
pb.update("Fetching revisions", 1, 2)
102
self._fetch_everything_for_search(search)
106
def _fetch_everything_for_search(self, search):
107
"""Fetch all data for the given set of revisions."""
108
# The first phase is "file". We pass the progress bar for it directly
109
# into item_keys_introduced_by, which has more information about how
110
# that phase is progressing than we do. Progress updates for the other
111
# phases are taken care of in this function.
112
# XXX: there should be a clear owner of the progress reporting. Perhaps
113
# item_keys_introduced_by should have a richer API than it does at the
114
# moment, so that it can feed the progress information back to this
116
if (self.from_repository._format.rich_root_data and
117
not self.to_repository._format.rich_root_data):
118
raise errors.IncompatibleRepositories(
119
self.from_repository, self.to_repository,
120
"different rich-root support")
121
pb = ui.ui_factory.nested_progress_bar()
123
pb.update("Get stream source")
124
source = self.from_repository._get_source(
125
self.to_repository._format)
126
stream = source.get_stream(search)
127
from_format = self.from_repository._format
128
pb.update("Inserting stream")
129
resume_tokens, missing_keys = self.sink.insert_stream(
130
stream, from_format, [])
131
if self.to_repository._fallback_repositories:
133
self._parent_inventories(search.get_keys()))
135
pb.update("Missing keys")
136
stream = source.get_stream_for_missing_keys(missing_keys)
137
pb.update("Inserting missing keys")
138
resume_tokens, missing_keys = self.sink.insert_stream(
139
stream, from_format, resume_tokens)
141
raise AssertionError(
142
"second push failed to complete a fetch %r." % (
145
raise AssertionError(
146
"second push failed to commit the fetch %r." % (
148
pb.update("Finishing stream")
153
def _revids_to_fetch(self):
154
"""Determines the exact revisions needed from self.from_repository to
155
install self._last_revision in self.to_repository.
157
If no revisions need to be fetched, then this just returns None.
159
if self._fetch_spec is not None:
160
return self._fetch_spec
161
mutter('fetch up to rev {%s}', self._last_revision)
162
if self._last_revision is NULL_REVISION:
163
# explicit limit of no revisions needed
165
return self.to_repository.search_missing_revision_ids(
166
self.from_repository, self._last_revision,
167
find_ghosts=self.find_ghosts)
169
def _parent_inventories(self, revision_ids):
170
# Find all the parent revisions referenced by the stream, but
171
# not present in the stream, and make sure we send their
173
parent_maps = self.to_repository.get_parent_map(revision_ids)
175
map(parents.update, parent_maps.itervalues())
176
parents.discard(NULL_REVISION)
177
parents.difference_update(revision_ids)
178
missing_keys = set(('inventories', rev_id) for rev_id in parents)
182
class Inter1and2Helper(object):
183
"""Helper for operations that convert data from model 1 and 2
185
This is for use by fetchers and converters.
188
def __init__(self, source):
191
:param source: The repository data comes from
195
def iter_rev_trees(self, revs):
196
"""Iterate through RevisionTrees efficiently.
198
Additionally, the inventory's revision_id is set if unset.
200
Trees are retrieved in batches of 100, and then yielded in the order
203
:param revs: A list of revision ids
205
# In case that revs is not a list.
208
for tree in self.source.revision_trees(revs[:100]):
209
if tree.inventory.revision_id is None:
210
tree.inventory.revision_id = tree.get_revision_id()
214
def _find_root_ids(self, revs, parent_map, graph):
216
for tree in self.iter_rev_trees(revs):
217
revision_id = tree.inventory.root.revision
218
root_id = tree.get_root_id()
219
revision_root[revision_id] = root_id
220
# Find out which parents we don't already know root ids for
222
for revision_parents in parent_map.itervalues():
223
parents.update(revision_parents)
224
parents.difference_update(revision_root.keys() + [NULL_REVISION])
225
# Limit to revisions present in the versionedfile
226
parents = graph.get_parent_map(parents).keys()
227
for tree in self.iter_rev_trees(parents):
228
root_id = tree.get_root_id()
229
revision_root[tree.get_revision_id()] = root_id
232
def generate_root_texts(self, revs):
233
"""Generate VersionedFiles for all root ids.
235
:param revs: the revisions to include
237
graph = self.source.get_graph()
238
parent_map = graph.get_parent_map(revs)
239
rev_order = tsort.topo_sort(parent_map)
240
rev_id_to_root_id = self._find_root_ids(revs, parent_map, graph)
241
root_id_order = [(rev_id_to_root_id[rev_id], rev_id) for rev_id in
243
# Guaranteed stable, this groups all the file id operations together
244
# retaining topological order within the revisions of a file id.
245
# File id splits and joins would invalidate this, but they don't exist
246
# yet, and are unlikely to in non-rich-root environments anyway.
247
root_id_order.sort(key=operator.itemgetter(0))
248
# Create a record stream containing the roots to create.
250
graph = _get_rich_root_heads_graph(self.source_repo, revs)
251
new_roots_stream = _new_root_data_stream(
252
root_id_order, rev_id_to_root_id, parent_map, self.source, graph)
253
return [('texts', new_roots_stream)]
256
def _get_rich_root_heads_graph(source_repo, revision_ids):
257
"""Get a Graph object suitable for asking heads() for new rich roots."""
258
st = static_tuple.StaticTuple
259
revision_keys = [st(r_id).intern() for r_id in revision_ids]
260
known_graph = source_repo.revisions.get_known_graph_ancestry(
262
return _mod_graph.GraphThunkIdsToKeys(known_graph)
265
def _new_root_data_stream(
266
root_keys_to_create, rev_id_to_root_id_map, parent_map, repo, graph=None):
267
"""Generate a texts substream of synthesised root entries.
269
Used in fetches that do rich-root upgrades.
271
:param root_keys_to_create: iterable of (root_id, rev_id) pairs describing
272
the root entries to create.
273
:param rev_id_to_root_id_map: dict of known rev_id -> root_id mappings for
274
calculating the parents. If a parent rev_id is not found here then it
275
will be recalculated.
276
:param parent_map: a parent map for all the revisions in
278
:param graph: a graph to use instead of repo.get_graph().
280
for root_key in root_keys_to_create:
281
root_id, rev_id = root_key
282
parent_keys = _parent_keys_for_root_version(
283
root_id, rev_id, rev_id_to_root_id_map, parent_map, repo, graph)
284
yield versionedfile.FulltextContentFactory(
285
root_key, parent_keys, None, '')
288
def _parent_keys_for_root_version(
289
root_id, rev_id, rev_id_to_root_id_map, parent_map, repo, graph=None):
290
"""Get the parent keys for a given root id.
292
A helper function for _new_root_data_stream.
294
# Include direct parents of the revision, but only if they used the same
295
# root_id and are heads.
296
rev_parents = parent_map[rev_id]
298
for parent_id in rev_parents:
299
if parent_id == NULL_REVISION:
301
if parent_id not in rev_id_to_root_id_map:
302
# We probably didn't read this revision, go spend the extra effort
305
tree = repo.revision_tree(parent_id)
306
except errors.NoSuchRevision:
307
# Ghost, fill out rev_id_to_root_id in case we encounter this
309
# But set parent_root_id to None since we don't really know
310
parent_root_id = None
312
parent_root_id = tree.get_root_id()
313
rev_id_to_root_id_map[parent_id] = None
315
# rev_id_to_root_id_map[parent_id] = parent_root_id
316
# memory consumption maybe?
318
parent_root_id = rev_id_to_root_id_map[parent_id]
319
if root_id == parent_root_id:
320
# With stacking we _might_ want to refer to a non-local revision,
321
# but this code path only applies when we have the full content
322
# available, so ghosts really are ghosts, not just the edge of
324
parent_ids.append(parent_id)
326
# root_id may be in the parent anyway.
328
tree = repo.revision_tree(parent_id)
329
except errors.NoSuchRevision:
330
# ghost, can't refer to it.
334
parent_ids.append(tree.inventory[root_id].revision)
335
except errors.NoSuchId:
338
# Drop non-head parents
340
graph = repo.get_graph()
341
heads = graph.heads(parent_ids)
343
for parent_id in parent_ids:
344
if parent_id in heads and parent_id not in selected_ids:
345
selected_ids.append(parent_id)
346
parent_keys = [(root_id, parent_id) for parent_id in selected_ids]