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
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(), """
40
from bzrlib.i18n import gettext
41
from bzrlib.revision import NULL_REVISION
42
from bzrlib.trace import mutter
45
class RepoFetcher(object):
46
"""Pull revisions and texts from one repository to another.
48
This should not be used directly, it's essential a object to encapsulate
49
the logic in InterRepository.fetch().
52
def __init__(self, to_repository, from_repository, last_revision=None,
53
find_ghosts=True, fetch_spec=None):
54
"""Create a repo fetcher.
56
:param last_revision: If set, try to limit to the data this revision
58
:param fetch_spec: A SearchResult specifying which revisions to fetch.
59
If set, this overrides last_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(gettext("Finding revisions"), 0, 2)
98
search_result = self._revids_to_fetch()
99
mutter('fetching: %s', search_result)
100
if search_result.is_empty():
102
pb.update(gettext("Fetching revisions"), 1, 2)
103
self._fetch_everything_for_search(search_result)
107
def _fetch_everything_for_search(self, search):
108
"""Fetch all data for the given set of revisions."""
109
# The first phase is "file". We pass the progress bar for it directly
110
# into item_keys_introduced_by, which has more information about how
111
# that phase is progressing than we do. Progress updates for the other
112
# phases are taken care of in this function.
113
# XXX: there should be a clear owner of the progress reporting. Perhaps
114
# item_keys_introduced_by should have a richer API than it does at the
115
# moment, so that it can feed the progress information back to this
117
if (self.from_repository._format.rich_root_data and
118
not self.to_repository._format.rich_root_data):
119
raise errors.IncompatibleRepositories(
120
self.from_repository, self.to_repository,
121
"different rich-root support")
122
pb = ui.ui_factory.nested_progress_bar()
124
pb.update("Get stream source")
125
source = self.from_repository._get_source(
126
self.to_repository._format)
127
stream = source.get_stream(search)
128
from_format = self.from_repository._format
129
pb.update("Inserting stream")
130
resume_tokens, missing_keys = self.sink.insert_stream(
131
stream, from_format, [])
133
pb.update("Missing keys")
134
stream = source.get_stream_for_missing_keys(missing_keys)
135
pb.update("Inserting missing keys")
136
resume_tokens, missing_keys = self.sink.insert_stream(
137
stream, from_format, resume_tokens)
139
raise AssertionError(
140
"second push failed to complete a fetch %r." % (
143
raise AssertionError(
144
"second push failed to commit the fetch %r." % (
146
pb.update("Finishing stream")
151
def _revids_to_fetch(self):
152
"""Determines the exact revisions needed from self.from_repository to
153
install self._last_revision in self.to_repository.
155
:returns: A SearchResult of some sort. (Possibly a
156
PendingAncestryResult, EmptySearchResult, etc.)
158
if self._fetch_spec is not None:
159
# The fetch spec is already a concrete search result.
160
return self._fetch_spec
161
elif self._last_revision == NULL_REVISION:
162
# fetch_spec is None + last_revision is null => empty fetch.
163
# explicit limit of no revisions needed
164
return _mod_graph.EmptySearchResult()
165
elif self._last_revision is not None:
166
return _mod_graph.NotInOtherForRevs(self.to_repository,
167
self.from_repository, [self._last_revision],
168
find_ghosts=self.find_ghosts).execute()
169
else: # self._last_revision is None:
170
return _mod_graph.EverythingNotInOther(self.to_repository,
171
self.from_repository,
172
find_ghosts=self.find_ghosts).execute()
175
class Inter1and2Helper(object):
176
"""Helper for operations that convert data from model 1 and 2
178
This is for use by fetchers and converters.
181
# This is a class variable so that the test suite can override it.
182
known_graph_threshold = 100
184
def __init__(self, source):
187
:param source: The repository data comes from
191
def iter_rev_trees(self, revs):
192
"""Iterate through RevisionTrees efficiently.
194
Additionally, the inventory's revision_id is set if unset.
196
Trees are retrieved in batches of 100, and then yielded in the order
199
:param revs: A list of revision ids
201
# In case that revs is not a list.
204
for tree in self.source.revision_trees(revs[:100]):
205
if tree.inventory.revision_id is None:
206
tree.inventory.revision_id = tree.get_revision_id()
210
def _find_root_ids(self, revs, parent_map, graph):
212
for tree in self.iter_rev_trees(revs):
213
revision_id = tree.inventory.root.revision
214
root_id = tree.get_root_id()
215
revision_root[revision_id] = root_id
216
# Find out which parents we don't already know root ids for
218
for revision_parents in parent_map.itervalues():
219
parents.update(revision_parents)
220
parents.difference_update(revision_root.keys() + [NULL_REVISION])
221
# Limit to revisions present in the versionedfile
222
parents = graph.get_parent_map(parents).keys()
223
for tree in self.iter_rev_trees(parents):
224
root_id = tree.get_root_id()
225
revision_root[tree.get_revision_id()] = root_id
228
def generate_root_texts(self, revs):
229
"""Generate VersionedFiles for all root ids.
231
:param revs: the revisions to include
233
graph = self.source.get_graph()
234
parent_map = graph.get_parent_map(revs)
235
rev_order = tsort.topo_sort(parent_map)
236
rev_id_to_root_id = self._find_root_ids(revs, parent_map, graph)
237
root_id_order = [(rev_id_to_root_id[rev_id], rev_id) for rev_id in
239
# Guaranteed stable, this groups all the file id operations together
240
# retaining topological order within the revisions of a file id.
241
# File id splits and joins would invalidate this, but they don't exist
242
# yet, and are unlikely to in non-rich-root environments anyway.
243
root_id_order.sort(key=operator.itemgetter(0))
244
# Create a record stream containing the roots to create.
245
if len(revs) > self.known_graph_threshold:
246
graph = self.source.get_known_graph_ancestry(revs)
247
new_roots_stream = _new_root_data_stream(
248
root_id_order, rev_id_to_root_id, parent_map, self.source, graph)
249
return [('texts', new_roots_stream)]
252
def _new_root_data_stream(
253
root_keys_to_create, rev_id_to_root_id_map, parent_map, repo, graph=None):
254
"""Generate a texts substream of synthesised root entries.
256
Used in fetches that do rich-root upgrades.
258
:param root_keys_to_create: iterable of (root_id, rev_id) pairs describing
259
the root entries to create.
260
:param rev_id_to_root_id_map: dict of known rev_id -> root_id mappings for
261
calculating the parents. If a parent rev_id is not found here then it
262
will be recalculated.
263
:param parent_map: a parent map for all the revisions in
265
:param graph: a graph to use instead of repo.get_graph().
267
for root_key in root_keys_to_create:
268
root_id, rev_id = root_key
269
parent_keys = _parent_keys_for_root_version(
270
root_id, rev_id, rev_id_to_root_id_map, parent_map, repo, graph)
271
yield versionedfile.FulltextContentFactory(
272
root_key, parent_keys, None, '')
275
def _parent_keys_for_root_version(
276
root_id, rev_id, rev_id_to_root_id_map, parent_map, repo, graph=None):
277
"""Get the parent keys for a given root id.
279
A helper function for _new_root_data_stream.
281
# Include direct parents of the revision, but only if they used the same
282
# root_id and are heads.
283
rev_parents = parent_map[rev_id]
285
for parent_id in rev_parents:
286
if parent_id == NULL_REVISION:
288
if parent_id not in rev_id_to_root_id_map:
289
# We probably didn't read this revision, go spend the extra effort
292
tree = repo.revision_tree(parent_id)
293
except errors.NoSuchRevision:
294
# Ghost, fill out rev_id_to_root_id in case we encounter this
296
# But set parent_root_id to None since we don't really know
297
parent_root_id = None
299
parent_root_id = tree.get_root_id()
300
rev_id_to_root_id_map[parent_id] = None
302
# rev_id_to_root_id_map[parent_id] = parent_root_id
303
# memory consumption maybe?
305
parent_root_id = rev_id_to_root_id_map[parent_id]
306
if root_id == parent_root_id:
307
# With stacking we _might_ want to refer to a non-local revision,
308
# but this code path only applies when we have the full content
309
# available, so ghosts really are ghosts, not just the edge of
311
parent_ids.append(parent_id)
313
# root_id may be in the parent anyway.
315
tree = repo.revision_tree(parent_id)
316
except errors.NoSuchRevision:
317
# ghost, can't refer to it.
321
parent_ids.append(tree.get_file_revision(root_id))
322
except errors.NoSuchId:
325
# Drop non-head parents
327
graph = repo.get_graph()
328
heads = graph.heads(parent_ids)
330
for parent_id in parent_ids:
331
if parent_id in heads and parent_id not in selected_ids:
332
selected_ids.append(parent_id)
333
parent_keys = [(root_id, parent_id) for parent_id in selected_ids]
337
class TargetRepoKinds(object):
338
"""An enum-like set of constants.
340
They are the possible values of FetchSpecFactory.target_repo_kinds.
343
PREEXISTING = 'preexisting'
348
class FetchSpecFactory(object):
349
"""A helper for building the best fetch spec for a sprout call.
351
Factors that go into determining the sort of fetch to perform:
352
* did the caller specify any revision IDs?
353
* did the caller specify a source branch (need to fetch its
354
heads_to_fetch(), usually the tip + tags)
355
* is there an existing target repo (don't need to refetch revs it
357
* target is stacked? (similar to pre-existing target repo: even if
358
the target itself is new don't want to refetch existing revs)
360
:ivar source_branch: the source branch if one specified, else None.
361
:ivar source_branch_stop_revision_id: fetch up to this revision of
362
source_branch, rather than its tip.
363
:ivar source_repo: the source repository if one found, else None.
364
:ivar target_repo: the target repository acquired by sprout.
365
:ivar target_repo_kind: one of the TargetRepoKinds constants.
369
self._explicit_rev_ids = set()
370
self.source_branch = None
371
self.source_branch_stop_revision_id = None
372
self.source_repo = None
373
self.target_repo = None
374
self.target_repo_kind = None
377
def add_revision_ids(self, revision_ids):
378
"""Add revision_ids to the set of revision_ids to be fetched."""
379
self._explicit_rev_ids.update(revision_ids)
381
def make_fetch_spec(self):
382
"""Build a SearchResult or PendingAncestryResult or etc."""
383
if self.target_repo_kind is None or self.source_repo is None:
384
raise AssertionError(
385
'Incomplete FetchSpecFactory: %r' % (self.__dict__,))
386
if len(self._explicit_rev_ids) == 0 and self.source_branch is None:
387
if self.limit is not None:
388
raise NotImplementedError(
389
"limit is only supported with a source branch set")
390
# Caller hasn't specified any revisions or source branch
391
if self.target_repo_kind == TargetRepoKinds.EMPTY:
392
return _mod_graph.EverythingResult(self.source_repo)
394
# We want everything not already in the target (or target's
396
return _mod_graph.EverythingNotInOther(
397
self.target_repo, self.source_repo).execute()
398
heads_to_fetch = set(self._explicit_rev_ids)
399
if self.source_branch is not None:
400
must_fetch, if_present_fetch = self.source_branch.heads_to_fetch()
401
if self.source_branch_stop_revision_id is not None:
402
# Replace the tip rev from must_fetch with the stop revision
403
# XXX: this might be wrong if the tip rev is also in the
404
# must_fetch set for other reasons (e.g. it's the tip of
405
# multiple loom threads?), but then it's pretty unclear what it
406
# should mean to specify a stop_revision in that case anyway.
407
must_fetch.discard(self.source_branch.last_revision())
408
must_fetch.add(self.source_branch_stop_revision_id)
409
heads_to_fetch.update(must_fetch)
411
if_present_fetch = set()
412
if self.target_repo_kind == TargetRepoKinds.EMPTY:
413
# PendingAncestryResult does not raise errors if a requested head
414
# is absent. Ideally it would support the
415
# required_ids/if_present_ids distinction, but in practice
416
# heads_to_fetch will almost certainly be present so this doesn't
418
all_heads = heads_to_fetch.union(if_present_fetch)
419
ret = _mod_graph.PendingAncestryResult(all_heads, self.source_repo)
420
if self.limit is not None:
421
graph = self.source_repo.get_graph()
422
topo_order = list(graph.iter_topo_order(ret.get_keys()))
423
result_set = topo_order[:self.limit]
424
ret = self.source_repo.revision_ids_to_search_result(result_set)
427
return _mod_graph.NotInOtherForRevs(self.target_repo, self.source_repo,
428
required_ids=heads_to_fetch, if_present_ids=if_present_fetch,
429
limit=self.limit).execute()
added
removed
bzrlib/fetch.py
