81
73
count_copied -- number of revisions copied
83
This should not be used directory, its essential a object to encapsulate
75
This should not be used directly, it's essential a object to encapsulate
84
76
the logic in InterRepository.fetch().
86
def __init__(self, to_repository, from_repository, last_revision=None, pb=None):
79
def __init__(self, to_repository, from_repository, last_revision=None, pb=None,
81
"""Create a repo fetcher.
83
:param find_ghosts: If True search the entire history for ghosts.
87
85
# result variables.
88
86
self.failed_revisions = []
89
87
self.count_copied = 0
90
if to_repository.control_files._transport.base == from_repository.control_files._transport.base:
91
# check that last_revision is in 'from' and then return a no-operation.
92
if last_revision not in (None, NULL_REVISION):
93
from_repository.get_revision(last_revision)
88
if to_repository.has_same_location(from_repository):
89
# repository.fetch should be taking care of this case.
90
raise errors.BzrError('RepoFetcher run '
91
'between two objects at the same location: '
92
'%r and %r' % (to_repository, from_repository))
95
93
self.to_repository = to_repository
96
94
self.from_repository = from_repository
97
95
# must not mutate self._last_revision as its potentially a shared instance
98
96
self._last_revision = last_revision
97
self.find_ghosts = find_ghosts
100
99
self.pb = bzrlib.ui.ui_factory.nested_progress_bar()
101
100
self.nested_pb = self.pb
120
128
This initialises all the needed variables, and then fetches the
121
129
requested revisions, finally clearing the progress bar.
123
self.to_weaves = self.to_repository.weave_store
124
self.to_control = self.to_repository.control_weaves
125
self.from_weaves = self.from_repository.weave_store
126
self.from_control = self.from_repository.control_weaves
127
131
self.count_total = 0
128
132
self.file_ids_names = {}
129
pp = ProgressPhase('Fetch phase', 4, self.pb)
133
pp = ProgressPhase('Transferring', 4, self.pb)
132
revs = self._revids_to_fetch()
136
self._fetch_weave_texts(revs)
138
self._fetch_inventory_weave(revs)
140
self._fetch_revision_texts(revs)
141
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
graph = self.from_repository.get_graph()
162
revs = list(graph.iter_topo_order(revs))
163
data_to_fetch = self.from_repository.item_keys_introduced_by(revs,
166
for knit_kind, file_id, revisions in data_to_fetch:
167
if knit_kind != phase:
169
# Make a new progress bar for this phase
172
pb = bzrlib.ui.ui_factory.nested_progress_bar()
173
if knit_kind == "file":
174
# Accumulate file texts
175
text_keys.extend([(file_id, revision) for revision in
177
elif knit_kind == "inventory":
178
# Now copy the file texts.
179
to_texts = self.to_repository.texts
180
from_texts = self.from_repository.texts
181
to_texts.insert_record_stream(from_texts.get_record_stream(
182
text_keys, self.to_repository._fetch_order,
183
self.to_repository._fetch_uses_deltas))
184
# Cause an error if a text occurs after we have done the
187
# Before we process the inventory we generate the root
188
# texts (if necessary) so that the inventories references
190
self._generate_root_texts(revs)
191
# NB: This currently reopens the inventory weave in source;
192
# using a single stream interface instead would avoid this.
193
self._fetch_inventory_weave(revs, pb)
194
elif knit_kind == "signatures":
195
# Nothing to do here; this will be taken care of when
196
# _fetch_revision_texts happens.
198
elif knit_kind == "revisions":
199
self._fetch_revision_texts(revs, pb)
201
raise AssertionError("Unknown knit kind %r" % knit_kind)
202
if self.to_repository._fetch_reconcile:
203
self.to_repository.reconcile()
207
self.count_copied += len(revs)
145
209
def _revids_to_fetch(self):
210
"""Determines the exact revisions needed from self.from_repository to
211
install self._last_revision in self.to_repository.
213
If no revisions need to be fetched, then this just returns None.
146
215
mutter('fetch up to rev {%s}', self._last_revision)
147
216
if self._last_revision is NULL_REVISION:
148
217
# explicit limit of no revisions needed
150
219
if (self._last_revision is not None and
151
220
self.to_repository.has_revision(self._last_revision)):
155
return self.to_repository.missing_revision_ids(self.from_repository,
157
except errors.NoSuchRevision:
223
return self.to_repository.search_missing_revision_ids(
224
self.from_repository, self._last_revision,
225
find_ghosts=self.find_ghosts)
226
except errors.NoSuchRevision, e:
158
227
raise InstallFailed([self._last_revision])
160
def _fetch_weave_texts(self, revs):
161
texts_pb = bzrlib.ui.ui_factory.nested_progress_bar()
163
# fileids_altered_by_revision_ids requires reading the inventory
164
# weave, we will need to read the inventory weave again when
165
# all this is done, so enable caching for that specific weave
166
inv_w = self.from_repository.get_inventory_weave()
168
file_ids = self.from_repository.fileids_altered_by_revision_ids(revs)
170
num_file_ids = len(file_ids)
171
for file_id, required_versions in file_ids.items():
172
texts_pb.update("fetch texts", count, num_file_ids)
174
to_weave = self.to_weaves.get_weave_or_empty(file_id,
175
self.to_repository.get_transaction())
176
from_weave = self.from_weaves.get_weave(file_id,
177
self.from_repository.get_transaction())
178
# we fetch all the texts, because texts do
179
# not reference anything, and its cheap enough
180
to_weave.join(from_weave, version_ids=required_versions)
181
# we don't need *all* of this data anymore, but we dont know
182
# what we do. This cache clearing will result in a new read
183
# of the knit data when we do the checkout, but probably we
184
# want to emit the needed data on the fly rather than at the
186
# the from weave should know not to cache data being joined,
187
# but its ok to ask it to clear.
188
from_weave.clear_cache()
189
to_weave.clear_cache()
193
def _fetch_inventory_weave(self, revs):
194
pb = bzrlib.ui.ui_factory.nested_progress_bar()
196
pb.update("fetch inventory", 0, 2)
197
to_weave = self.to_control.get_weave('inventory',
198
self.to_repository.get_transaction())
200
child_pb = bzrlib.ui.ui_factory.nested_progress_bar()
202
# just merge, this is optimisable and its means we don't
203
# copy unreferenced data such as not-needed inventories.
204
pb.update("fetch inventory", 1, 3)
205
from_weave = self.from_repository.get_inventory_weave()
206
pb.update("fetch inventory", 2, 3)
207
# we fetch only the referenced inventories because we do not
208
# know for unselected inventories whether all their required
209
# texts are present in the other repository - it could be
211
to_weave.join(from_weave, pb=child_pb, msg='merge inventory',
213
from_weave.clear_cache()
220
class GenericRepoFetcher(RepoFetcher):
221
"""This is a generic repo to repo fetcher.
223
This makes minimal assumptions about repo layout and contents.
224
It triggers a reconciliation after fetching to ensure integrity.
227
def _fetch_revision_texts(self, revs):
228
"""Fetch revision object texts"""
229
rev_pb = bzrlib.ui.ui_factory.nested_progress_bar()
231
to_txn = self.to_transaction = self.to_repository.get_transaction()
234
to_store = self.to_repository._revision_store
236
pb = bzrlib.ui.ui_factory.nested_progress_bar()
238
pb.update('copying revisions', count, total)
240
sig_text = self.from_repository.get_signature_text(rev)
241
to_store.add_revision_signature_text(rev, sig_text, to_txn)
242
except errors.NoSuchRevision:
245
to_store.add_revision(self.from_repository.get_revision(rev),
250
# fixup inventory if needed:
251
# this is expensive because we have no inverse index to current ghosts.
252
# but on local disk its a few seconds and sftp push is already insane.
254
# FIXME: repository should inform if this is needed.
255
self.to_repository.reconcile()
260
class KnitRepoFetcher(RepoFetcher):
261
"""This is a knit format repository specific fetcher.
263
This differs from the GenericRepoFetcher by not doing a
264
reconciliation after copying, and using knit joining to
268
def _fetch_revision_texts(self, revs):
229
def _fetch_inventory_weave(self, revs, pb):
230
pb.update("fetch inventory", 0, 2)
231
to_weave = self.to_repository.inventories
232
child_pb = bzrlib.ui.ui_factory.nested_progress_bar()
234
# just merge, this is optimisable and its means we don't
235
# copy unreferenced data such as not-needed inventories.
236
pb.update("fetch inventory", 1, 3)
237
from_weave = self.from_repository.inventories
238
pb.update("fetch inventory", 2, 3)
239
# we fetch only the referenced inventories because we do not
240
# know for unselected inventories whether all their required
241
# texts are present in the other repository - it could be
243
to_weave.insert_record_stream(from_weave.get_record_stream(
244
[(rev_id,) for rev_id in revs],
245
self.to_repository._fetch_order,
246
self.to_repository._fetch_uses_deltas))
250
def _fetch_revision_texts(self, revs, pb):
269
251
# may need to be a InterRevisionStore call here.
270
from_transaction = self.from_repository.get_transaction()
271
to_transaction = self.to_repository.get_transaction()
272
to_sf = self.to_repository._revision_store.get_signature_file(
274
from_sf = self.from_repository._revision_store.get_signature_file(
276
to_sf.join(from_sf, version_ids=revs, ignore_missing=True)
277
to_rf = self.to_repository._revision_store.get_revision_file(
279
from_rf = self.from_repository._revision_store.get_revision_file(
281
to_rf.join(from_rf, version_ids=revs)
252
to_sf = self.to_repository.signatures
253
from_sf = self.from_repository.signatures
254
# A missing signature is just skipped.
255
to_sf.insert_record_stream(filter_absent(from_sf.get_record_stream(
256
[(rev_id,) for rev_id in revs],
257
self.to_repository._fetch_order,
258
self.to_repository._fetch_uses_deltas)))
259
self._fetch_just_revision_texts(revs)
261
def _fetch_just_revision_texts(self, version_ids):
262
to_rf = self.to_repository.revisions
263
from_rf = self.from_repository.revisions
264
to_rf.insert_record_stream(from_rf.get_record_stream(
265
[(rev_id,) for rev_id in version_ids],
266
self.to_repository._fetch_order,
267
self.to_repository._fetch_uses_deltas))
269
def _generate_root_texts(self, revs):
270
"""This will be called by __fetch between fetching weave texts and
271
fetching the inventory weave.
273
Subclasses should override this if they need to generate root texts
274
after fetching weave texts.
284
279
class Inter1and2Helper(object):
314
311
revs = revs[100:]
313
def _find_root_ids(self, revs, parent_map, graph):
315
planned_versions = {}
316
for tree in self.iter_rev_trees(revs):
317
revision_id = tree.inventory.root.revision
318
root_id = tree.get_root_id()
319
planned_versions.setdefault(root_id, []).append(revision_id)
320
revision_root[revision_id] = root_id
321
# Find out which parents we don't already know root ids for
323
for revision_parents in parent_map.itervalues():
324
parents.update(revision_parents)
325
parents.difference_update(revision_root.keys() + [NULL_REVISION])
326
# Limit to revisions present in the versionedfile
327
parents = graph.get_parent_map(parents).keys()
328
for tree in self.iter_rev_trees(parents):
329
root_id = tree.get_root_id()
330
revision_root[tree.get_revision_id()] = root_id
331
return revision_root, planned_versions
316
333
def generate_root_texts(self, revs):
317
334
"""Generate VersionedFiles for all root ids.
319
336
:param revs: the revisions to include
321
inventory_weave = self.source.get_inventory_weave()
324
to_store = self.target.weave_store
325
for tree in self.iter_rev_trees(revs):
326
revision_id = tree.inventory.root.revision
327
root_id = tree.inventory.root.file_id
328
parents = inventory_weave.get_parents(revision_id)
329
if root_id not in versionedfile:
330
versionedfile[root_id] = to_store.get_weave_or_empty(root_id,
331
self.target.get_transaction())
332
parent_texts[root_id] = versionedfile[root_id].add_lines(
333
revision_id, parents, [], parent_texts)
338
to_texts = self.target.texts
339
graph = self.source.get_graph()
340
parent_map = graph.get_parent_map(revs)
341
rev_order = topo_sort(parent_map)
342
rev_id_to_root_id, root_id_to_rev_ids = self._find_root_ids(
343
revs, parent_map, graph)
344
root_id_order = [(rev_id_to_root_id[rev_id], rev_id) for rev_id in
346
# Guaranteed stable, this groups all the file id operations together
347
# retaining topological order within the revisions of a file id.
348
# File id splits and joins would invalidate this, but they don't exist
349
# yet, and are unlikely to in non-rich-root environments anyway.
350
root_id_order.sort(key=operator.itemgetter(0))
351
# Create a record stream containing the roots to create.
353
for key in root_id_order:
354
root_id, rev_id = key
355
rev_parents = parent_map[rev_id]
356
# We drop revision parents with different file-ids, because
357
# that represents a rename of the root to a different location
358
# - its not actually a parent for us. (We could look for that
359
# file id in the revision tree at considerably more expense,
360
# but for now this is sufficient (and reconcile will catch and
361
# correct this anyway).
362
# When a parent revision is a ghost, we guess that its root id
363
# was unchanged (rather than trimming it from the parent list).
364
parent_keys = tuple((root_id, parent) for parent in rev_parents
365
if parent != NULL_REVISION and
366
rev_id_to_root_id.get(parent, root_id) == root_id)
367
yield FulltextContentFactory(key, parent_keys, None, '')
368
to_texts.insert_record_stream(yield_roots())
335
370
def regenerate_inventory(self, revs):
336
371
"""Generate a new inventory versionedfile in target, convertin data.
339
374
stored in the target (reserializing it in a different format).
340
375
:param revs: The revisions to include
342
inventory_weave = self.source.get_inventory_weave()
343
377
for tree in self.iter_rev_trees(revs):
344
parents = inventory_weave.get_parents(tree.get_revision_id())
378
parents = tree.get_parent_ids()
345
379
self.target.add_inventory(tree.get_revision_id(), tree.inventory,
349
class Model1toKnit2Fetcher(GenericRepoFetcher):
382
def fetch_revisions(self, revision_ids):
383
for revision in self.source.get_revisions(revision_ids):
384
self.target.add_revision(revision.revision_id, revision)
387
class Model1toKnit2Fetcher(RepoFetcher):
350
388
"""Fetch from a Model1 repository into a Knit2 repository
352
def __init__(self, to_repository, from_repository, last_revision=None,
390
def __init__(self, to_repository, from_repository, last_revision=None,
391
pb=None, find_ghosts=True):
354
392
self.helper = Inter1and2Helper(from_repository, to_repository)
355
GenericRepoFetcher.__init__(self, to_repository, from_repository,
393
RepoFetcher.__init__(self, to_repository, from_repository,
394
last_revision, pb, find_ghosts)
358
def _fetch_weave_texts(self, revs):
359
GenericRepoFetcher._fetch_weave_texts(self, revs)
360
# Now generate a weave for the tree root
396
def _generate_root_texts(self, revs):
361
397
self.helper.generate_root_texts(revs)
363
def _fetch_inventory_weave(self, revs):
399
def _fetch_inventory_weave(self, revs, pb):
364
400
self.helper.regenerate_inventory(revs)
367
class Knit1to2Fetcher(KnitRepoFetcher):
402
def _fetch_revision_texts(self, revs, pb):
403
"""Fetch revision object texts"""
407
pb.update('copying revisions', count, total)
409
sig_text = self.from_repository.get_signature_text(rev)
410
self.to_repository.add_signature_text(rev, sig_text)
411
except errors.NoSuchRevision:
414
self._copy_revision(rev)
417
def _copy_revision(self, rev):
418
self.helper.fetch_revisions([rev])
421
class Knit1to2Fetcher(RepoFetcher):
368
422
"""Fetch from a Knit1 repository into a Knit2 repository"""
370
def __init__(self, to_repository, from_repository, last_revision=None,
424
def __init__(self, to_repository, from_repository, last_revision=None,
425
pb=None, find_ghosts=True):
372
426
self.helper = Inter1and2Helper(from_repository, to_repository)
373
KnitRepoFetcher.__init__(self, to_repository, from_repository,
427
RepoFetcher.__init__(self, to_repository, from_repository,
428
last_revision, pb, find_ghosts)
376
def _fetch_weave_texts(self, revs):
377
KnitRepoFetcher._fetch_weave_texts(self, revs)
378
# Now generate a weave for the tree root
430
def _generate_root_texts(self, revs):
379
431
self.helper.generate_root_texts(revs)
381
def _fetch_inventory_weave(self, revs):
433
def _fetch_inventory_weave(self, revs, pb):
382
434
self.helper.regenerate_inventory(revs)
385
class Fetcher(object):
386
"""Backwards compatibility glue for branch.fetch()."""
388
@deprecated_method(zero_eight)
389
def __init__(self, to_branch, from_branch, last_revision=None, pb=None):
390
"""Please see branch.fetch()."""
391
to_branch.fetch(from_branch, last_revision, pb)
436
def _fetch_just_revision_texts(self, version_ids):
437
self.helper.fetch_revisions(version_ids)