51
58
def __init__(self, to_repository, from_repository, last_revision=None,
52
find_ghosts=True, fetch_spec=None):
59
find_ghosts=True, fetch_spec=None):
53
60
"""Create a repo fetcher.
55
62
:param last_revision: If set, try to limit to the data this revision
64
:param fetch_spec: A SearchResult specifying which revisions to fetch.
65
If set, this overrides last_revision.
57
66
:param find_ghosts: If True search the entire history for ghosts.
59
68
# repository.fetch has the responsibility for short-circuiting
88
97
# assert not missing
89
98
self.count_total = 0
90
99
self.file_ids_names = {}
91
pb = ui.ui_factory.nested_progress_bar()
92
pb.show_pct = pb.show_count = False
94
pb.update("Finding revisions", 0, 2)
95
search = self._revids_to_fetch()
100
with ui.ui_factory.nested_progress_bar() as pb:
101
pb.show_pct = pb.show_count = False
102
pb.update(gettext("Finding revisions"), 0, 2)
103
search_result = self._revids_to_fetch()
104
mutter('fetching: %s', str(search_result))
105
if search_result.is_empty():
98
pb.update("Fetching revisions", 1, 2)
99
self._fetch_everything_for_search(search)
107
pb.update(gettext("Fetching revisions"), 1, 2)
108
self._fetch_everything_for_search(search_result)
103
110
def _fetch_everything_for_search(self, search):
104
111
"""Fetch all data for the given set of revisions."""
111
118
# moment, so that it can feed the progress information back to this
113
120
if (self.from_repository._format.rich_root_data and
114
not self.to_repository._format.rich_root_data):
121
not self.to_repository._format.rich_root_data):
115
122
raise errors.IncompatibleRepositories(
116
123
self.from_repository, self.to_repository,
117
124
"different rich-root support")
118
pb = ui.ui_factory.nested_progress_bar()
125
with ui.ui_factory.nested_progress_bar() as pb:
120
126
pb.update("Get stream source")
121
127
source = self.from_repository._get_source(
122
128
self.to_repository._format)
145
148
pb.update("Finishing stream")
146
149
self.sink.finished()
150
151
def _revids_to_fetch(self):
151
152
"""Determines the exact revisions needed from self.from_repository to
152
153
install self._last_revision in self.to_repository.
154
If no revisions need to be fetched, then this just returns None.
155
:returns: A SearchResult of some sort. (Possibly a
156
PendingAncestryResult, EmptySearchResult, etc.)
156
158
if self._fetch_spec is not None:
159
# The fetch spec is already a concrete search result.
157
160
return self._fetch_spec
158
mutter('fetch up to rev {%s}', self._last_revision)
159
if self._last_revision is NULL_REVISION:
161
elif self._last_revision == NULL_REVISION:
162
# fetch_spec is None + last_revision is null => empty fetch.
160
163
# explicit limit of no revisions needed
162
return self.to_repository.search_missing_revision_ids(
163
self.from_repository, self._last_revision,
164
find_ghosts=self.find_ghosts)
166
def _parent_inventories(self, revision_ids):
167
# Find all the parent revisions referenced by the stream, but
168
# not present in the stream, and make sure we send their
170
parent_maps = self.to_repository.get_parent_map(revision_ids)
172
map(parents.update, parent_maps.itervalues())
173
parents.discard(NULL_REVISION)
174
parents.difference_update(revision_ids)
175
missing_keys = set(('inventories', rev_id) for rev_id in parents)
164
return vf_search.EmptySearchResult()
165
elif self._last_revision is not None:
166
return vf_search.NotInOtherForRevs(self.to_repository,
167
self.from_repository, [
168
self._last_revision],
169
find_ghosts=self.find_ghosts).execute()
170
else: # self._last_revision is None:
171
return vf_search.EverythingNotInOther(self.to_repository,
172
self.from_repository,
173
find_ghosts=self.find_ghosts).execute()
179
176
class Inter1and2Helper(object):
203
203
revs = list(revs)
205
205
for tree in self.source.revision_trees(revs[:100]):
206
if tree.inventory.revision_id is None:
207
tree.inventory.revision_id = tree.get_revision_id()
206
if tree.root_inventory.revision_id is None:
207
tree.root_inventory.revision_id = tree.get_revision_id()
209
209
revs = revs[100:]
211
211
def _find_root_ids(self, revs, parent_map, graph):
212
212
revision_root = {}
213
213
for tree in self.iter_rev_trees(revs):
214
revision_id = tree.inventory.root.revision
215
214
root_id = tree.get_root_id()
215
revision_id = tree.get_file_revision(u'')
216
216
revision_root[revision_id] = root_id
217
217
# Find out which parents we don't already know root ids for
219
for revision_parents in parent_map.itervalues():
220
parents.update(revision_parents)
221
parents.difference_update(revision_root.keys() + [NULL_REVISION])
218
parents = set(viewvalues(parent_map))
219
parents.difference_update(revision_root)
220
parents.discard(NULL_REVISION)
222
221
# Limit to revisions present in the versionedfile
223
parents = graph.get_parent_map(parents).keys()
222
parents = graph.get_parent_map(parents)
224
223
for tree in self.iter_rev_trees(parents):
225
224
root_id = tree.get_root_id()
226
225
revision_root[tree.get_revision_id()] = root_id
236
235
rev_order = tsort.topo_sort(parent_map)
237
236
rev_id_to_root_id = self._find_root_ids(revs, parent_map, graph)
238
237
root_id_order = [(rev_id_to_root_id[rev_id], rev_id) for rev_id in
240
239
# Guaranteed stable, this groups all the file id operations together
241
240
# retaining topological order within the revisions of a file id.
242
241
# File id splits and joins would invalidate this, but they don't exist
243
242
# yet, and are unlikely to in non-rich-root environments anyway.
244
243
root_id_order.sort(key=operator.itemgetter(0))
245
244
# Create a record stream containing the roots to create.
247
# XXX: not covered by tests, should have a flag to always run
248
# this. -- mbp 20100129
249
graph = self.source_repo.get_known_graph_ancestry(revs)
245
if len(revs) > self.known_graph_threshold:
246
graph = self.source.get_known_graph_ancestry(revs)
250
247
new_roots_stream = _new_root_data_stream(
251
248
root_id_order, rev_id_to_root_id, parent_map, self.source, graph)
252
249
return [('texts', new_roots_stream)]
255
def _get_rich_root_heads_graph(source_repo, revision_ids):
256
"""Get a Graph object suitable for asking heads() for new rich roots."""
260
252
def _new_root_data_stream(
261
root_keys_to_create, rev_id_to_root_id_map, parent_map, repo, graph=None):
253
root_keys_to_create, rev_id_to_root_id_map, parent_map, repo, graph=None):
262
254
"""Generate a texts substream of synthesised root entries.
264
256
Used in fetches that do rich-root upgrades.
266
258
:param root_keys_to_create: iterable of (root_id, rev_id) pairs describing
267
259
the root entries to create.
268
260
:param rev_id_to_root_id_map: dict of known rev_id -> root_id mappings for
277
269
parent_keys = _parent_keys_for_root_version(
278
270
root_id, rev_id, rev_id_to_root_id_map, parent_map, repo, graph)
279
271
yield versionedfile.FulltextContentFactory(
280
root_key, parent_keys, None, '')
272
root_key, parent_keys, None, b'')
283
275
def _parent_keys_for_root_version(
284
root_id, rev_id, rev_id_to_root_id_map, parent_map, repo, graph=None):
276
root_id, rev_id, rev_id_to_root_id_map, parent_map, repo, graph=None):
285
277
"""Get the parent keys for a given root id.
287
279
A helper function for _new_root_data_stream.
289
281
# Include direct parents of the revision, but only if they used the same
340
333
selected_ids.append(parent_id)
341
334
parent_keys = [(root_id, parent_id) for parent_id in selected_ids]
342
335
return parent_keys
338
class TargetRepoKinds(object):
339
"""An enum-like set of constants.
341
They are the possible values of FetchSpecFactory.target_repo_kinds.
344
PREEXISTING = 'preexisting'
349
class FetchSpecFactory(object):
350
"""A helper for building the best fetch spec for a sprout call.
352
Factors that go into determining the sort of fetch to perform:
353
* did the caller specify any revision IDs?
354
* did the caller specify a source branch (need to fetch its
355
heads_to_fetch(), usually the tip + tags)
356
* is there an existing target repo (don't need to refetch revs it
358
* target is stacked? (similar to pre-existing target repo: even if
359
the target itself is new don't want to refetch existing revs)
361
:ivar source_branch: the source branch if one specified, else None.
362
:ivar source_branch_stop_revision_id: fetch up to this revision of
363
source_branch, rather than its tip.
364
:ivar source_repo: the source repository if one found, else None.
365
:ivar target_repo: the target repository acquired by sprout.
366
:ivar target_repo_kind: one of the TargetRepoKinds constants.
370
self._explicit_rev_ids = set()
371
self.source_branch = None
372
self.source_branch_stop_revision_id = None
373
self.source_repo = None
374
self.target_repo = None
375
self.target_repo_kind = None
378
def add_revision_ids(self, revision_ids):
379
"""Add revision_ids to the set of revision_ids to be fetched."""
380
self._explicit_rev_ids.update(revision_ids)
382
def make_fetch_spec(self):
383
"""Build a SearchResult or PendingAncestryResult or etc."""
384
if self.target_repo_kind is None or self.source_repo is None:
385
raise AssertionError(
386
'Incomplete FetchSpecFactory: %r' % (self.__dict__,))
387
if len(self._explicit_rev_ids) == 0 and self.source_branch is None:
388
if self.limit is not None:
389
raise NotImplementedError(
390
"limit is only supported with a source branch set")
391
# Caller hasn't specified any revisions or source branch
392
if self.target_repo_kind == TargetRepoKinds.EMPTY:
393
return vf_search.EverythingResult(self.source_repo)
395
# We want everything not already in the target (or target's
397
return vf_search.EverythingNotInOther(
398
self.target_repo, self.source_repo).execute()
399
heads_to_fetch = set(self._explicit_rev_ids)
400
if self.source_branch is not None:
401
must_fetch, if_present_fetch = self.source_branch.heads_to_fetch()
402
if self.source_branch_stop_revision_id is not None:
403
# Replace the tip rev from must_fetch with the stop revision
404
# XXX: this might be wrong if the tip rev is also in the
405
# must_fetch set for other reasons (e.g. it's the tip of
406
# multiple loom threads?), but then it's pretty unclear what it
407
# should mean to specify a stop_revision in that case anyway.
408
must_fetch.discard(self.source_branch.last_revision())
409
must_fetch.add(self.source_branch_stop_revision_id)
410
heads_to_fetch.update(must_fetch)
412
if_present_fetch = set()
413
if self.target_repo_kind == TargetRepoKinds.EMPTY:
414
# PendingAncestryResult does not raise errors if a requested head
415
# is absent. Ideally it would support the
416
# required_ids/if_present_ids distinction, but in practice
417
# heads_to_fetch will almost certainly be present so this doesn't
419
all_heads = heads_to_fetch.union(if_present_fetch)
420
ret = vf_search.PendingAncestryResult(all_heads, self.source_repo)
421
if self.limit is not None:
422
graph = self.source_repo.get_graph()
423
topo_order = list(graph.iter_topo_order(ret.get_keys()))
424
result_set = topo_order[:self.limit]
425
ret = self.source_repo.revision_ids_to_search_result(
429
return vf_search.NotInOtherForRevs(self.target_repo, self.source_repo,
430
required_ids=heads_to_fetch, if_present_ids=if_present_fetch,
431
limit=self.limit).execute()