1
# Copyright (C) 2005-2011 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
17
"""Copying of history from one branch to another.
19
The basic plan is that every branch knows the history of everything
20
that has merged into it. As the first step of a merge, pull, or
21
branch operation we copy history from the source into the destination
25
from __future__ import absolute_import
29
from .lazy_import import lazy_import
30
lazy_import(globals(), """
41
from .i18n import gettext
42
from .revision import NULL_REVISION
46
from .trace import mutter
49
class RepoFetcher(object):
50
"""Pull revisions and texts from one repository to another.
52
This should not be used directly, it's essential a object to encapsulate
53
the logic in InterRepository.fetch().
56
def __init__(self, to_repository, from_repository, last_revision=None,
57
find_ghosts=True, fetch_spec=None):
58
"""Create a repo fetcher.
60
:param last_revision: If set, try to limit to the data this revision
62
:param fetch_spec: A SearchResult specifying which revisions to fetch.
63
If set, this overrides last_revision.
64
:param find_ghosts: If True search the entire history for ghosts.
66
# repository.fetch has the responsibility for short-circuiting
67
# attempts to copy between a repository and itself.
68
self.to_repository = to_repository
69
self.from_repository = from_repository
70
self.sink = to_repository._get_sink()
71
# must not mutate self._last_revision as its potentially a shared instance
72
self._last_revision = last_revision
73
self._fetch_spec = fetch_spec
74
self.find_ghosts = find_ghosts
75
self.from_repository.lock_read()
76
mutter("Using fetch logic to copy between %s(%s) and %s(%s)",
77
self.from_repository, self.from_repository._format,
78
self.to_repository, self.to_repository._format)
82
self.from_repository.unlock()
85
"""Primary worker function.
87
This initialises all the needed variables, and then fetches the
88
requested revisions, finally clearing the progress bar.
90
# Roughly this is what we're aiming for fetch to become:
92
# missing = self.sink.insert_stream(self.source.get_stream(search))
94
# missing = self.sink.insert_stream(self.source.get_items(missing))
97
self.file_ids_names = {}
98
pb = ui.ui_factory.nested_progress_bar()
99
pb.show_pct = pb.show_count = False
101
pb.update(gettext("Finding revisions"), 0, 2)
102
search_result = self._revids_to_fetch()
103
mutter('fetching: %s', search_result)
104
if search_result.is_empty():
106
pb.update(gettext("Fetching revisions"), 1, 2)
107
self._fetch_everything_for_search(search_result)
111
def _fetch_everything_for_search(self, search):
112
"""Fetch all data for the given set of revisions."""
113
# The first phase is "file". We pass the progress bar for it directly
114
# into item_keys_introduced_by, which has more information about how
115
# that phase is progressing than we do. Progress updates for the other
116
# phases are taken care of in this function.
117
# XXX: there should be a clear owner of the progress reporting. Perhaps
118
# item_keys_introduced_by should have a richer API than it does at the
119
# moment, so that it can feed the progress information back to this
121
if (self.from_repository._format.rich_root_data and
122
not self.to_repository._format.rich_root_data):
123
raise errors.IncompatibleRepositories(
124
self.from_repository, self.to_repository,
125
"different rich-root support")
126
pb = ui.ui_factory.nested_progress_bar()
128
pb.update("Get stream source")
129
source = self.from_repository._get_source(
130
self.to_repository._format)
131
stream = source.get_stream(search)
132
from_format = self.from_repository._format
133
pb.update("Inserting stream")
134
resume_tokens, missing_keys = self.sink.insert_stream(
135
stream, from_format, [])
137
pb.update("Missing keys")
138
stream = source.get_stream_for_missing_keys(missing_keys)
139
pb.update("Inserting missing keys")
140
resume_tokens, missing_keys = self.sink.insert_stream(
141
stream, from_format, resume_tokens)
143
raise AssertionError(
144
"second push failed to complete a fetch %r." % (
147
raise AssertionError(
148
"second push failed to commit the fetch %r." % (
150
pb.update("Finishing stream")
155
def _revids_to_fetch(self):
156
"""Determines the exact revisions needed from self.from_repository to
157
install self._last_revision in self.to_repository.
159
:returns: A SearchResult of some sort. (Possibly a
160
PendingAncestryResult, EmptySearchResult, etc.)
162
if self._fetch_spec is not None:
163
# The fetch spec is already a concrete search result.
164
return self._fetch_spec
165
elif self._last_revision == NULL_REVISION:
166
# fetch_spec is None + last_revision is null => empty fetch.
167
# explicit limit of no revisions needed
168
return vf_search.EmptySearchResult()
169
elif self._last_revision is not None:
170
return vf_search.NotInOtherForRevs(self.to_repository,
171
self.from_repository, [self._last_revision],
172
find_ghosts=self.find_ghosts).execute()
173
else: # self._last_revision is None:
174
return vf_search.EverythingNotInOther(self.to_repository,
175
self.from_repository,
176
find_ghosts=self.find_ghosts).execute()
179
class Inter1and2Helper(object):
180
"""Helper for operations that convert data from model 1 and 2
182
This is for use by fetchers and converters.
185
# This is a class variable so that the test suite can override it.
186
known_graph_threshold = 100
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.root_inventory.revision_id is None:
210
tree.root_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
root_id = tree.get_root_id()
218
revision_id = tree.get_file_revision(root_id, u"")
219
revision_root[revision_id] = root_id
220
# Find out which parents we don't already know root ids for
221
parents = set(viewvalues(parent_map))
222
parents.difference_update(revision_root)
223
parents.discard(NULL_REVISION)
224
# Limit to revisions present in the versionedfile
225
parents = graph.get_parent_map(parents)
226
for tree in self.iter_rev_trees(parents):
227
root_id = tree.get_root_id()
228
revision_root[tree.get_revision_id()] = root_id
231
def generate_root_texts(self, revs):
232
"""Generate VersionedFiles for all root ids.
234
:param revs: the revisions to include
236
graph = self.source.get_graph()
237
parent_map = graph.get_parent_map(revs)
238
rev_order = tsort.topo_sort(parent_map)
239
rev_id_to_root_id = self._find_root_ids(revs, parent_map, graph)
240
root_id_order = [(rev_id_to_root_id[rev_id], rev_id) for rev_id in
242
# Guaranteed stable, this groups all the file id operations together
243
# retaining topological order within the revisions of a file id.
244
# File id splits and joins would invalidate this, but they don't exist
245
# yet, and are unlikely to in non-rich-root environments anyway.
246
root_id_order.sort(key=operator.itemgetter(0))
247
# Create a record stream containing the roots to create.
248
if len(revs) > self.known_graph_threshold:
249
graph = self.source.get_known_graph_ancestry(revs)
250
new_roots_stream = _new_root_data_stream(
251
root_id_order, rev_id_to_root_id, parent_map, self.source, graph)
252
return [('texts', new_roots_stream)]
255
def _new_root_data_stream(
256
root_keys_to_create, rev_id_to_root_id_map, parent_map, repo, graph=None):
257
"""Generate a texts substream of synthesised root entries.
259
Used in fetches that do rich-root upgrades.
261
:param root_keys_to_create: iterable of (root_id, rev_id) pairs describing
262
the root entries to create.
263
:param rev_id_to_root_id_map: dict of known rev_id -> root_id mappings for
264
calculating the parents. If a parent rev_id is not found here then it
265
will be recalculated.
266
:param parent_map: a parent map for all the revisions in
268
:param graph: a graph to use instead of repo.get_graph().
270
for root_key in root_keys_to_create:
271
root_id, rev_id = root_key
272
parent_keys = _parent_keys_for_root_version(
273
root_id, rev_id, rev_id_to_root_id_map, parent_map, repo, graph)
274
yield versionedfile.FulltextContentFactory(
275
root_key, parent_keys, None, '')
278
def _parent_keys_for_root_version(
279
root_id, rev_id, rev_id_to_root_id_map, parent_map, repo, graph=None):
280
"""Get the parent keys for a given root id.
282
A helper function for _new_root_data_stream.
284
# Include direct parents of the revision, but only if they used the same
285
# root_id and are heads.
286
rev_parents = parent_map[rev_id]
288
for parent_id in rev_parents:
289
if parent_id == NULL_REVISION:
291
if parent_id not in rev_id_to_root_id_map:
292
# We probably didn't read this revision, go spend the extra effort
295
tree = repo.revision_tree(parent_id)
296
except errors.NoSuchRevision:
297
# Ghost, fill out rev_id_to_root_id in case we encounter this
299
# But set parent_root_id to None since we don't really know
300
parent_root_id = None
302
parent_root_id = tree.get_root_id()
303
rev_id_to_root_id_map[parent_id] = None
305
# rev_id_to_root_id_map[parent_id] = parent_root_id
306
# memory consumption maybe?
308
parent_root_id = rev_id_to_root_id_map[parent_id]
309
if root_id == parent_root_id:
310
# With stacking we _might_ want to refer to a non-local revision,
311
# but this code path only applies when we have the full content
312
# available, so ghosts really are ghosts, not just the edge of
314
parent_ids.append(parent_id)
316
# root_id may be in the parent anyway.
318
tree = repo.revision_tree(parent_id)
319
except errors.NoSuchRevision:
320
# ghost, can't refer to it.
324
parent_ids.append(tree.get_file_revision(root_id))
325
except errors.NoSuchId:
328
# Drop non-head parents
330
graph = repo.get_graph()
331
heads = graph.heads(parent_ids)
333
for parent_id in parent_ids:
334
if parent_id in heads and parent_id not in selected_ids:
335
selected_ids.append(parent_id)
336
parent_keys = [(root_id, parent_id) for parent_id in selected_ids]
340
class TargetRepoKinds(object):
341
"""An enum-like set of constants.
343
They are the possible values of FetchSpecFactory.target_repo_kinds.
346
PREEXISTING = 'preexisting'
351
class FetchSpecFactory(object):
352
"""A helper for building the best fetch spec for a sprout call.
354
Factors that go into determining the sort of fetch to perform:
355
* did the caller specify any revision IDs?
356
* did the caller specify a source branch (need to fetch its
357
heads_to_fetch(), usually the tip + tags)
358
* is there an existing target repo (don't need to refetch revs it
360
* target is stacked? (similar to pre-existing target repo: even if
361
the target itself is new don't want to refetch existing revs)
363
:ivar source_branch: the source branch if one specified, else None.
364
:ivar source_branch_stop_revision_id: fetch up to this revision of
365
source_branch, rather than its tip.
366
:ivar source_repo: the source repository if one found, else None.
367
:ivar target_repo: the target repository acquired by sprout.
368
:ivar target_repo_kind: one of the TargetRepoKinds constants.
372
self._explicit_rev_ids = set()
373
self.source_branch = None
374
self.source_branch_stop_revision_id = None
375
self.source_repo = None
376
self.target_repo = None
377
self.target_repo_kind = None
380
def add_revision_ids(self, revision_ids):
381
"""Add revision_ids to the set of revision_ids to be fetched."""
382
self._explicit_rev_ids.update(revision_ids)
384
def make_fetch_spec(self):
385
"""Build a SearchResult or PendingAncestryResult or etc."""
386
if self.target_repo_kind is None or self.source_repo is None:
387
raise AssertionError(
388
'Incomplete FetchSpecFactory: %r' % (self.__dict__,))
389
if len(self._explicit_rev_ids) == 0 and self.source_branch is None:
390
if self.limit is not None:
391
raise NotImplementedError(
392
"limit is only supported with a source branch set")
393
# Caller hasn't specified any revisions or source branch
394
if self.target_repo_kind == TargetRepoKinds.EMPTY:
395
return vf_search.EverythingResult(self.source_repo)
397
# We want everything not already in the target (or target's
399
return vf_search.EverythingNotInOther(
400
self.target_repo, self.source_repo).execute()
401
heads_to_fetch = set(self._explicit_rev_ids)
402
if self.source_branch is not None:
403
must_fetch, if_present_fetch = self.source_branch.heads_to_fetch()
404
if self.source_branch_stop_revision_id is not None:
405
# Replace the tip rev from must_fetch with the stop revision
406
# XXX: this might be wrong if the tip rev is also in the
407
# must_fetch set for other reasons (e.g. it's the tip of
408
# multiple loom threads?), but then it's pretty unclear what it
409
# should mean to specify a stop_revision in that case anyway.
410
must_fetch.discard(self.source_branch.last_revision())
411
must_fetch.add(self.source_branch_stop_revision_id)
412
heads_to_fetch.update(must_fetch)
414
if_present_fetch = set()
415
if self.target_repo_kind == TargetRepoKinds.EMPTY:
416
# PendingAncestryResult does not raise errors if a requested head
417
# is absent. Ideally it would support the
418
# required_ids/if_present_ids distinction, but in practice
419
# heads_to_fetch will almost certainly be present so this doesn't
421
all_heads = heads_to_fetch.union(if_present_fetch)
422
ret = vf_search.PendingAncestryResult(all_heads, self.source_repo)
423
if self.limit is not None:
424
graph = self.source_repo.get_graph()
425
topo_order = list(graph.iter_topo_order(ret.get_keys()))
426
result_set = topo_order[:self.limit]
427
ret = self.source_repo.revision_ids_to_search_result(result_set)
430
return vf_search.NotInOtherForRevs(self.target_repo, self.source_repo,
431
required_ids=heads_to_fetch, if_present_ids=if_present_fetch,
432
limit=self.limit).execute()