145
148
pb.update("Finishing stream")
146
149
self.sink.finished()
148
153
def _revids_to_fetch(self):
149
154
"""Determines the exact revisions needed from self.from_repository to
150
155
install self._last_revision in self.to_repository.
152
:returns: A SearchResult of some sort. (Possibly a
153
PendingAncestryResult, EmptySearchResult, etc.)
157
If no revisions need to be fetched, then this just returns None.
155
159
if self._fetch_spec is not None:
156
# The fetch spec is already a concrete search result.
157
160
return self._fetch_spec
158
elif self._last_revision == NULL_REVISION:
159
# fetch_spec is None + last_revision is null => empty fetch.
161
mutter('fetch up to rev {%s}', self._last_revision)
162
if self._last_revision is NULL_REVISION:
160
163
# explicit limit of no revisions needed
161
return vf_search.EmptySearchResult()
162
elif self._last_revision is not None:
163
return vf_search.NotInOtherForRevs(self.to_repository,
164
self.from_repository, [
165
self._last_revision],
166
find_ghosts=self.find_ghosts).execute()
167
else: # self._last_revision is None:
168
return vf_search.EverythingNotInOther(self.to_repository,
169
self.from_repository,
170
find_ghosts=self.find_ghosts).execute()
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)
173
182
class Inter1and2Helper(object):
230
239
graph = self.source.get_graph()
231
240
parent_map = graph.get_parent_map(revs)
232
rev_order = tsort.topo_sort(parent_map)
233
rev_id_to_root_id = self._find_root_ids(revs, parent_map, graph)
241
rev_order = topo_sort(parent_map)
242
rev_id_to_root_id, root_id_to_rev_ids = self._find_root_ids(
243
revs, parent_map, graph)
234
244
root_id_order = [(rev_id_to_root_id[rev_id], rev_id) for rev_id in
236
246
# Guaranteed stable, this groups all the file id operations together
237
247
# retaining topological order within the revisions of a file id.
238
248
# File id splits and joins would invalidate this, but they don't exist
239
249
# yet, and are unlikely to in non-rich-root environments anyway.
240
250
root_id_order.sort(key=operator.itemgetter(0))
241
251
# Create a record stream containing the roots to create.
242
if len(revs) > self.known_graph_threshold:
243
graph = self.source.get_known_graph_ancestry(revs)
244
new_roots_stream = _new_root_data_stream(
245
root_id_order, rev_id_to_root_id, parent_map, self.source, graph)
246
return [('texts', new_roots_stream)]
249
def _new_root_data_stream(
250
root_keys_to_create, rev_id_to_root_id_map, parent_map, repo, graph=None):
251
"""Generate a texts substream of synthesised root entries.
253
Used in fetches that do rich-root upgrades.
255
:param root_keys_to_create: iterable of (root_id, rev_id) pairs describing
256
the root entries to create.
257
:param rev_id_to_root_id_map: dict of known rev_id -> root_id mappings for
258
calculating the parents. If a parent rev_id is not found here then it
259
will be recalculated.
260
:param parent_map: a parent map for all the revisions in
262
:param graph: a graph to use instead of repo.get_graph().
264
for root_key in root_keys_to_create:
265
root_id, rev_id = root_key
266
parent_keys = _parent_keys_for_root_version(
267
root_id, rev_id, rev_id_to_root_id_map, parent_map, repo, graph)
268
yield versionedfile.ChunkedContentFactory(
269
root_key, parent_keys, None, [])
272
def _parent_keys_for_root_version(
273
root_id, rev_id, rev_id_to_root_id_map, parent_map, repo, graph=None):
274
"""Get the parent keys for a given root id.
276
A helper function for _new_root_data_stream.
278
# Include direct parents of the revision, but only if they used the same
279
# root_id and are heads.
280
rev_parents = parent_map[rev_id]
282
for parent_id in rev_parents:
283
if parent_id == NULL_REVISION:
285
if parent_id not in rev_id_to_root_id_map:
286
# We probably didn't read this revision, go spend the extra effort
289
tree = repo.revision_tree(parent_id)
290
except errors.NoSuchRevision:
291
# Ghost, fill out rev_id_to_root_id in case we encounter this
293
# But set parent_root_id to None since we don't really know
294
parent_root_id = None
296
parent_root_id = tree.path2id('')
297
rev_id_to_root_id_map[parent_id] = None
299
# rev_id_to_root_id_map[parent_id] = parent_root_id
300
# memory consumption maybe?
302
parent_root_id = rev_id_to_root_id_map[parent_id]
303
if root_id == parent_root_id:
304
# With stacking we _might_ want to refer to a non-local revision,
305
# but this code path only applies when we have the full content
306
# available, so ghosts really are ghosts, not just the edge of
308
parent_ids.append(parent_id)
310
# root_id may be in the parent anyway.
312
tree = repo.revision_tree(parent_id)
313
except errors.NoSuchRevision:
314
# ghost, can't refer to it.
319
tree.get_file_revision(
320
tree.id2path(root_id, recurse='none')))
321
except errors.NoSuchId:
324
# Drop non-head parents
326
graph = repo.get_graph()
327
heads = graph.heads(parent_ids)
329
for parent_id in parent_ids:
330
if parent_id in heads and parent_id not in selected_ids:
331
selected_ids.append(parent_id)
332
parent_keys = [(root_id, parent_id) for parent_id in selected_ids]
336
class TargetRepoKinds(object):
337
"""An enum-like set of constants.
339
They are the possible values of FetchSpecFactory.target_repo_kinds.
342
PREEXISTING = 'preexisting'
347
class FetchSpecFactory(object):
348
"""A helper for building the best fetch spec for a sprout call.
350
Factors that go into determining the sort of fetch to perform:
351
* did the caller specify any revision IDs?
352
* did the caller specify a source branch (need to fetch its
353
heads_to_fetch(), usually the tip + tags)
354
* is there an existing target repo (don't need to refetch revs it
356
* target is stacked? (similar to pre-existing target repo: even if
357
the target itself is new don't want to refetch existing revs)
359
:ivar source_branch: the source branch if one specified, else None.
360
:ivar source_branch_stop_revision_id: fetch up to this revision of
361
source_branch, rather than its tip.
362
:ivar source_repo: the source repository if one found, else None.
363
:ivar target_repo: the target repository acquired by sprout.
364
:ivar target_repo_kind: one of the TargetRepoKinds constants.
368
self._explicit_rev_ids = set()
369
self.source_branch = None
370
self.source_branch_stop_revision_id = None
371
self.source_repo = None
372
self.target_repo = None
373
self.target_repo_kind = None
376
def add_revision_ids(self, revision_ids):
377
"""Add revision_ids to the set of revision_ids to be fetched."""
378
self._explicit_rev_ids.update(revision_ids)
380
def make_fetch_spec(self):
381
"""Build a SearchResult or PendingAncestryResult or etc."""
382
if self.target_repo_kind is None or self.source_repo is None:
383
raise AssertionError(
384
'Incomplete FetchSpecFactory: %r' % (self.__dict__,))
385
if len(self._explicit_rev_ids) == 0 and self.source_branch is None:
386
if self.limit is not None:
387
raise NotImplementedError(
388
"limit is only supported with a source branch set")
389
# Caller hasn't specified any revisions or source branch
390
if self.target_repo_kind == TargetRepoKinds.EMPTY:
391
return vf_search.EverythingResult(self.source_repo)
393
# We want everything not already in the target (or target's
395
return vf_search.EverythingNotInOther(
396
self.target_repo, self.source_repo).execute()
397
heads_to_fetch = set(self._explicit_rev_ids)
398
if self.source_branch is not None:
399
must_fetch, if_present_fetch = self.source_branch.heads_to_fetch()
400
if self.source_branch_stop_revision_id is not None:
401
# Replace the tip rev from must_fetch with the stop revision
402
# XXX: this might be wrong if the tip rev is also in the
403
# must_fetch set for other reasons (e.g. it's the tip of
404
# multiple loom threads?), but then it's pretty unclear what it
405
# should mean to specify a stop_revision in that case anyway.
406
must_fetch.discard(self.source_branch.last_revision())
407
must_fetch.add(self.source_branch_stop_revision_id)
408
heads_to_fetch.update(must_fetch)
410
if_present_fetch = set()
411
if self.target_repo_kind == TargetRepoKinds.EMPTY:
412
# PendingAncestryResult does not raise errors if a requested head
413
# is absent. Ideally it would support the
414
# required_ids/if_present_ids distinction, but in practice
415
# heads_to_fetch will almost certainly be present so this doesn't
417
all_heads = heads_to_fetch.union(if_present_fetch)
418
ret = vf_search.PendingAncestryResult(all_heads, self.source_repo)
419
if self.limit is not None:
420
graph = self.source_repo.get_graph()
421
topo_order = list(graph.iter_topo_order(ret.get_keys()))
422
result_set = topo_order[:self.limit]
423
ret = self.source_repo.revision_ids_to_search_result(
427
return vf_search.NotInOtherForRevs(self.target_repo, self.source_repo,
428
required_ids=heads_to_fetch, if_present_ids=if_present_fetch,
429
limit=self.limit).execute()
253
for key in root_id_order:
254
root_id, rev_id = key
255
rev_parents = parent_map[rev_id]
256
# We drop revision parents with different file-ids, because
257
# that represents a rename of the root to a different location
258
# - its not actually a parent for us. (We could look for that
259
# file id in the revision tree at considerably more expense,
260
# but for now this is sufficient (and reconcile will catch and
261
# correct this anyway).
262
# When a parent revision is a ghost, we guess that its root id
263
# was unchanged (rather than trimming it from the parent list).
264
parent_keys = tuple((root_id, parent) for parent in rev_parents
265
if parent != NULL_REVISION and
266
rev_id_to_root_id.get(parent, root_id) == root_id)
267
yield FulltextContentFactory(key, parent_keys, None, '')
268
return [('texts', yield_roots())]