79
71
count_copied -- number of revisions copied
81
This should not be used directory, its essential a object to encapsulate
73
This should not be used directly, it's essential a object to encapsulate
82
74
the logic in InterRepository.fetch().
84
def __init__(self, to_repository, from_repository, last_revision=None, pb=None):
77
def __init__(self, to_repository, from_repository, last_revision=None, pb=None,
79
"""Create a repo fetcher.
81
:param find_ghosts: If True search the entire history for ghosts.
85
83
# result variables.
86
84
self.failed_revisions = []
87
85
self.count_copied = 0
88
if to_repository.control_files._transport.base == from_repository.control_files._transport.base:
89
# check that last_revision is in 'from' and then return a no-operation.
90
if last_revision not in (None, NULL_REVISION):
91
from_repository.get_revision(last_revision)
86
if to_repository.has_same_location(from_repository):
87
# repository.fetch should be taking care of this case.
88
raise errors.BzrError('RepoFetcher run '
89
'between two objects at the same location: '
90
'%r and %r' % (to_repository, from_repository))
93
91
self.to_repository = to_repository
94
92
self.from_repository = from_repository
95
93
# must not mutate self._last_revision as its potentially a shared instance
96
94
self._last_revision = last_revision
95
self.find_ghosts = find_ghosts
98
97
self.pb = bzrlib.ui.ui_factory.nested_progress_bar()
99
98
self.nested_pb = self.pb
119
127
requested revisions, finally clearing the progress bar.
121
129
self.to_weaves = self.to_repository.weave_store
122
self.to_control = self.to_repository.control_weaves
123
130
self.from_weaves = self.from_repository.weave_store
124
self.from_control = self.from_repository.control_weaves
125
131
self.count_total = 0
126
132
self.file_ids_names = {}
127
pp = ProgressPhase('fetch phase', 4, self.pb)
133
pp = ProgressPhase('Transferring', 4, self.pb)
129
revs = self._revids_to_fetch()
133
self._fetch_weave_texts(revs)
135
self._fetch_inventory_weave(revs)
137
self._fetch_revision_texts(revs)
138
self.count_copied += len(revs)
136
search = self._revids_to_fetch()
139
if getattr(self, '_fetch_everything_for_search', None) is not None:
140
self._fetch_everything_for_search(search, pp)
142
# backward compatibility
143
self._fetch_everything_for_revisions(search.get_keys, pp)
147
def _fetch_everything_for_search(self, search, pp):
148
"""Fetch all data for the given set of revisions."""
149
# The first phase is "file". We pass the progress bar for it directly
150
# into item_keys_introduced_by, which has more information about how
151
# that phase is progressing than we do. Progress updates for the other
152
# phases are taken care of in this function.
153
# XXX: there should be a clear owner of the progress reporting. Perhaps
154
# item_keys_introduced_by should have a richer API than it does at the
155
# moment, so that it can feed the progress information back to this
158
pb = bzrlib.ui.ui_factory.nested_progress_bar()
160
revs = search.get_keys()
161
data_to_fetch = self.from_repository.item_keys_introduced_by(revs, pb)
162
for knit_kind, file_id, revisions in data_to_fetch:
163
if knit_kind != phase:
165
# Make a new progress bar for this phase
168
pb = bzrlib.ui.ui_factory.nested_progress_bar()
169
if knit_kind == "file":
170
self._fetch_weave_text(file_id, revisions)
171
elif knit_kind == "inventory":
172
# Before we process the inventory we generate the root
173
# texts (if necessary) so that the inventories references
175
self._generate_root_texts(revs)
176
# NB: This currently reopens the inventory weave in source;
177
# using a full get_data_stream instead would avoid this.
178
self._fetch_inventory_weave(revs, pb)
179
elif knit_kind == "signatures":
180
# Nothing to do here; this will be taken care of when
181
# _fetch_revision_texts happens.
183
elif knit_kind == "revisions":
184
self._fetch_revision_texts(revs, pb)
186
raise AssertionError("Unknown knit kind %r" % knit_kind)
190
self.count_copied += len(revs)
142
192
def _revids_to_fetch(self):
193
"""Determines the exact revisions needed from self.from_repository to
194
install self._last_revision in self.to_repository.
196
If no revisions need to be fetched, then this just returns None.
143
198
mutter('fetch up to rev {%s}', self._last_revision)
144
199
if self._last_revision is NULL_REVISION:
145
200
# explicit limit of no revisions needed
147
if (self._last_revision != None and
202
if (self._last_revision is not None and
148
203
self.to_repository.has_revision(self._last_revision)):
152
return self.to_repository.missing_revision_ids(self.from_repository,
206
return self.to_repository.search_missing_revision_ids(
207
self.from_repository, self._last_revision,
208
find_ghosts=self.find_ghosts)
154
209
except errors.NoSuchRevision:
155
210
raise InstallFailed([self._last_revision])
157
def _fetch_weave_texts(self, revs):
158
texts_pb = bzrlib.ui.ui_factory.nested_progress_bar()
160
file_ids = self.from_repository.fileid_involved_by_set(revs)
162
num_file_ids = len(file_ids)
163
for file_id in file_ids:
164
texts_pb.update("fetch texts", count, num_file_ids)
167
to_weave = self.to_weaves.get_weave(file_id,
168
self.to_repository.get_transaction())
169
except errors.NoSuchFile:
170
# destination is empty, just copy it.
171
# this copies all the texts, which is useful and
172
# on per-file basis quite cheap.
173
self.to_weaves.copy_multi(
177
self.from_repository.get_transaction(),
178
self.to_repository.get_transaction())
180
# destination has contents, must merge
181
from_weave = self.from_weaves.get_weave(file_id,
182
self.from_repository.get_transaction())
183
# we fetch all the texts, because texts do
184
# not reference anything, and its cheap enough
185
to_weave.join(from_weave)
189
def _fetch_inventory_weave(self, revs):
190
pb = bzrlib.ui.ui_factory.nested_progress_bar()
192
pb.update("fetch inventory", 0, 2)
193
to_weave = self.to_control.get_weave('inventory',
194
self.to_repository.get_transaction())
196
child_pb = bzrlib.ui.ui_factory.nested_progress_bar()
198
# just merge, this is optimisable and its means we dont
199
# copy unreferenced data such as not-needed inventories.
200
pb.update("fetch inventory", 1, 3)
201
from_weave = self.from_repository.get_inventory_weave()
202
pb.update("fetch inventory", 2, 3)
203
# we fetch only the referenced inventories because we do not
204
# know for unselected inventories whether all their required
205
# texts are present in the other repository - it could be
207
to_weave.join(from_weave, pb=child_pb, msg='merge inventory',
212
def _fetch_weave_text(self, file_id, required_versions):
213
to_weave = self.to_weaves.get_weave_or_empty(file_id,
214
self.to_repository.get_transaction())
215
from_weave = self.from_weaves.get_weave(file_id,
216
self.from_repository.get_transaction())
217
# we fetch all the texts, because texts do
218
# not reference anything, and its cheap enough
219
to_weave.join(from_weave, version_ids=required_versions)
221
def _fetch_inventory_weave(self, revs, pb):
222
pb.update("fetch inventory", 0, 2)
223
to_weave = self.to_repository.get_inventory_weave()
224
child_pb = bzrlib.ui.ui_factory.nested_progress_bar()
226
# just merge, this is optimisable and its means we don't
227
# copy unreferenced data such as not-needed inventories.
228
pb.update("fetch inventory", 1, 3)
229
from_weave = self.from_repository.get_inventory_weave()
230
pb.update("fetch inventory", 2, 3)
231
# we fetch only the referenced inventories because we do not
232
# know for unselected inventories whether all their required
233
# texts are present in the other repository - it could be
235
to_weave.join(from_weave, pb=child_pb, msg='merge inventory',
240
def _generate_root_texts(self, revs):
241
"""This will be called by __fetch between fetching weave texts and
242
fetching the inventory weave.
244
Subclasses should override this if they need to generate root texts
245
after fetching weave texts.
215
250
class GenericRepoFetcher(RepoFetcher):
219
254
It triggers a reconciliation after fetching to ensure integrity.
222
def _fetch_revision_texts(self, revs):
257
def _fetch_revision_texts(self, revs, pb):
223
258
"""Fetch revision object texts"""
224
rev_pb = bzrlib.ui.ui_factory.nested_progress_bar()
226
to_txn = self.to_transaction = self.to_repository.get_transaction()
229
to_store = self.to_repository._revision_store
231
pb = bzrlib.ui.ui_factory.nested_progress_bar()
233
pb.update('copying revisions', count, total)
235
sig_text = self.from_repository.get_signature_text(rev)
236
to_store.add_revision_signature_text(rev, sig_text, to_txn)
237
except errors.NoSuchRevision:
240
to_store.add_revision(self.from_repository.get_revision(rev),
245
# fixup inventory if needed:
246
# this is expensive because we have no inverse index to current ghosts.
247
# but on local disk its a few seconds and sftp push is already insane.
249
# FIXME: repository should inform if this is needed.
250
self.to_repository.reconcile()
259
to_txn = self.to_transaction = self.to_repository.get_transaction()
262
to_store = self.to_repository._revision_store
264
pb.update('copying revisions', count, total)
266
sig_text = self.from_repository.get_signature_text(rev)
267
to_store.add_revision_signature_text(rev, sig_text, to_txn)
268
except errors.NoSuchRevision:
271
to_store.add_revision(self.from_repository.get_revision(rev),
274
# fixup inventory if needed:
275
# this is expensive because we have no inverse index to current ghosts.
276
# but on local disk its a few seconds and sftp push is already insane.
278
# FIXME: repository should inform if this is needed.
279
self.to_repository.reconcile()
255
282
class KnitRepoFetcher(RepoFetcher):
276
303
to_rf.join(from_rf, version_ids=revs)
279
class Fetcher(object):
280
"""Backwards compatability glue for branch.fetch()."""
282
@deprecated_method(zero_eight)
283
def __init__(self, to_branch, from_branch, last_revision=None, pb=None):
284
"""Please see branch.fetch()."""
285
to_branch.fetch(from_branch, last_revision, pb)
306
class Inter1and2Helper(object):
307
"""Helper for operations that convert data from model 1 and 2
309
This is for use by fetchers and converters.
312
def __init__(self, source, target):
315
:param source: The repository data comes from
316
:param target: The repository data goes to
321
def iter_rev_trees(self, revs):
322
"""Iterate through RevisionTrees efficiently.
324
Additionally, the inventory's revision_id is set if unset.
326
Trees are retrieved in batches of 100, and then yielded in the order
329
:param revs: A list of revision ids
331
# In case that revs is not a list.
334
for tree in self.source.revision_trees(revs[:100]):
335
if tree.inventory.revision_id is None:
336
tree.inventory.revision_id = tree.get_revision_id()
340
def generate_root_texts(self, revs):
341
"""Generate VersionedFiles for all root ids.
343
:param revs: the revisions to include
345
inventory_weave = self.source.get_inventory_weave()
348
to_store = self.target.weave_store
349
parent_map = self.source.get_graph().get_parent_map(revs)
350
for tree in self.iter_rev_trees(revs):
351
revision_id = tree.inventory.root.revision
352
root_id = tree.get_root_id()
353
parents = parent_map[revision_id]
354
if parents[0] == NULL_REVISION:
356
if root_id not in versionedfile:
357
versionedfile[root_id] = to_store.get_weave_or_empty(root_id,
358
self.target.get_transaction())
359
_, _, parent_texts[root_id] = versionedfile[root_id].add_lines(
360
revision_id, parents, [], parent_texts)
362
def regenerate_inventory(self, revs):
363
"""Generate a new inventory versionedfile in target, convertin data.
365
The inventory is retrieved from the source, (deserializing it), and
366
stored in the target (reserializing it in a different format).
367
:param revs: The revisions to include
369
for tree in self.iter_rev_trees(revs):
370
parents = tree.get_parent_ids()
371
self.target.add_inventory(tree.get_revision_id(), tree.inventory,
375
class Model1toKnit2Fetcher(GenericRepoFetcher):
376
"""Fetch from a Model1 repository into a Knit2 repository
378
def __init__(self, to_repository, from_repository, last_revision=None,
379
pb=None, find_ghosts=True):
380
self.helper = Inter1and2Helper(from_repository, to_repository)
381
GenericRepoFetcher.__init__(self, to_repository, from_repository,
382
last_revision, pb, find_ghosts)
384
def _generate_root_texts(self, revs):
385
self.helper.generate_root_texts(revs)
387
def _fetch_inventory_weave(self, revs, pb):
388
self.helper.regenerate_inventory(revs)
391
class Knit1to2Fetcher(KnitRepoFetcher):
392
"""Fetch from a Knit1 repository into a Knit2 repository"""
394
def __init__(self, to_repository, from_repository, last_revision=None,
395
pb=None, find_ghosts=True):
396
self.helper = Inter1and2Helper(from_repository, to_repository)
397
KnitRepoFetcher.__init__(self, to_repository, from_repository,
398
last_revision, pb, find_ghosts)
400
def _generate_root_texts(self, revs):
401
self.helper.generate_root_texts(revs)
403
def _fetch_inventory_weave(self, revs, pb):
404
self.helper.regenerate_inventory(revs)
407
class RemoteToOtherFetcher(GenericRepoFetcher):
409
def _fetch_everything_for_search(self, search, pp):
410
data_stream = self.from_repository.get_data_stream_for_search(search)
411
self.to_repository.insert_data_stream(data_stream)