1
# Copyright (C) 2005, 2006, 2008 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., 59 Temple Place, Suite 330, Boston, MA 02111-1307 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
25
The copying is done in a slightly complicated order. We don't want to
26
add a revision to the store until everything it refers to is also
27
stored, so that if a revision is present we can totally recreate it.
28
However, we can't know what files are included in a revision until we
29
read its inventory. So we query the inventory store of the source for
30
the ids we need, and then pull those ids and then return to the inventories.
36
import bzrlib.errors as errors
37
from bzrlib.errors import InstallFailed
38
from bzrlib.progress import ProgressPhase
39
from bzrlib.revision import is_null, NULL_REVISION
40
from bzrlib.symbol_versioning import (deprecated_function,
43
from bzrlib.tsort import topo_sort
44
from bzrlib.trace import mutter
46
from bzrlib.versionedfile import filter_absent, FulltextContentFactory
48
# TODO: Avoid repeatedly opening weaves so many times.
50
# XXX: This doesn't handle ghost (not present in branch) revisions at
51
# all yet. I'm not sure they really should be supported.
53
# NOTE: This doesn't copy revisions which may be present but not
54
# merged into the last revision. I'm not sure we want to do that.
56
# - get a list of revisions that need to be pulled in
57
# - for each one, pull in that revision file
58
# and get the inventory, and store the inventory with right
60
# - and get the ancestry, and store that with right parents too
61
# - and keep a note of all file ids and version seen
62
# - then go through all files; for each one get the weave,
63
# and add in all file versions
66
class RepoFetcher(object):
67
"""Pull revisions and texts from one repository to another.
70
if set, try to limit to the data this revision references.
73
count_copied -- number of revisions copied
75
This should not be used directly, it's essential a object to encapsulate
76
the logic in InterRepository.fetch().
79
def __init__(self, to_repository, from_repository, last_revision=None, pb=None,
80
find_ghosts=True, _write_group_acquired_callable=None):
81
"""Create a repo fetcher.
83
:param find_ghosts: If True search the entire history for ghosts.
84
:param _write_group_acquired_callable: Don't use; this parameter only
85
exists to facilitate a hack done in InterPackRepo.fetch. We would
86
like to remove this parameter.
89
self.failed_revisions = []
91
if to_repository.has_same_location(from_repository):
92
# repository.fetch should be taking care of this case.
93
raise errors.BzrError('RepoFetcher run '
94
'between two objects at the same location: '
95
'%r and %r' % (to_repository, from_repository))
96
self.to_repository = to_repository
97
self.from_repository = from_repository
98
# must not mutate self._last_revision as its potentially a shared instance
99
self._last_revision = last_revision
100
self.find_ghosts = find_ghosts
101
self._write_group_acquired_callable = _write_group_acquired_callable
103
self.pb = bzrlib.ui.ui_factory.nested_progress_bar()
104
self.nested_pb = self.pb
107
self.nested_pb = None
108
self.from_repository.lock_read()
110
self.to_repository.lock_write()
112
self.to_repository.start_write_group()
114
if self._write_group_acquired_callable is not None:
115
# Used by InterPackRepo.fetch to set_write_cache_size
117
self._write_group_acquired_callable()
120
self.to_repository.abort_write_group(suppress_errors=True)
123
self.to_repository.commit_write_group()
126
if self.nested_pb is not None:
127
self.nested_pb.finished()
129
self.to_repository.unlock()
131
self.from_repository.unlock()
134
"""Primary worker function.
136
This initialises all the needed variables, and then fetches the
137
requested revisions, finally clearing the progress bar.
140
self.file_ids_names = {}
141
pp = ProgressPhase('Transferring', 4, self.pb)
144
search = self._revids_to_fetch()
147
if getattr(self, '_fetch_everything_for_search', None) is not None:
148
self._fetch_everything_for_search(search, pp)
150
# backward compatibility
151
self._fetch_everything_for_revisions(search.get_keys, pp)
155
def _fetch_everything_for_search(self, search, pp):
156
"""Fetch all data for the given set of revisions."""
157
# The first phase is "file". We pass the progress bar for it directly
158
# into item_keys_introduced_by, which has more information about how
159
# that phase is progressing than we do. Progress updates for the other
160
# phases are taken care of in this function.
161
# XXX: there should be a clear owner of the progress reporting. Perhaps
162
# item_keys_introduced_by should have a richer API than it does at the
163
# moment, so that it can feed the progress information back to this
166
pb = bzrlib.ui.ui_factory.nested_progress_bar()
168
revs = search.get_keys()
169
graph = self.from_repository.get_graph()
170
revs = list(graph.iter_topo_order(revs))
171
data_to_fetch = self.from_repository.item_keys_introduced_by(revs,
174
for knit_kind, file_id, revisions in data_to_fetch:
175
if knit_kind != phase:
177
# Make a new progress bar for this phase
180
pb = bzrlib.ui.ui_factory.nested_progress_bar()
181
if knit_kind == "file":
182
# Accumulate file texts
183
text_keys.extend([(file_id, revision) for revision in
185
elif knit_kind == "inventory":
186
# Now copy the file texts.
187
to_texts = self.to_repository.texts
188
from_texts = self.from_repository.texts
189
to_texts.insert_record_stream(from_texts.get_record_stream(
190
text_keys, self.to_repository._fetch_order,
191
not self.to_repository._fetch_uses_deltas))
192
# Cause an error if a text occurs after we have done the
195
# Before we process the inventory we generate the root
196
# texts (if necessary) so that the inventories references
198
self._generate_root_texts(revs)
199
# NB: This currently reopens the inventory weave in source;
200
# using a single stream interface instead would avoid this.
201
self._fetch_inventory_weave(revs, pb)
202
elif knit_kind == "signatures":
203
# Nothing to do here; this will be taken care of when
204
# _fetch_revision_texts happens.
206
elif knit_kind == "revisions":
207
self._fetch_revision_texts(revs, pb)
209
raise AssertionError("Unknown knit kind %r" % knit_kind)
210
if self.to_repository._fetch_reconcile:
211
self.to_repository.reconcile()
215
self.count_copied += len(revs)
217
def _revids_to_fetch(self):
218
"""Determines the exact revisions needed from self.from_repository to
219
install self._last_revision in self.to_repository.
221
If no revisions need to be fetched, then this just returns None.
223
mutter('fetch up to rev {%s}', self._last_revision)
224
if self._last_revision is NULL_REVISION:
225
# explicit limit of no revisions needed
227
if (self._last_revision is not None and
228
self.to_repository.has_revision(self._last_revision)):
231
return self.to_repository.search_missing_revision_ids(
232
self.from_repository, self._last_revision,
233
find_ghosts=self.find_ghosts)
234
except errors.NoSuchRevision, e:
235
raise InstallFailed([self._last_revision])
237
def _fetch_inventory_weave(self, revs, pb):
238
pb.update("fetch inventory", 0, 2)
239
to_weave = self.to_repository.inventories
240
child_pb = bzrlib.ui.ui_factory.nested_progress_bar()
242
# just merge, this is optimisable and its means we don't
243
# copy unreferenced data such as not-needed inventories.
244
pb.update("fetch inventory", 1, 3)
245
from_weave = self.from_repository.inventories
246
pb.update("fetch inventory", 2, 3)
247
# we fetch only the referenced inventories because we do not
248
# know for unselected inventories whether all their required
249
# texts are present in the other repository - it could be
251
to_weave.insert_record_stream(from_weave.get_record_stream(
252
[(rev_id,) for rev_id in revs],
253
self.to_repository._fetch_order,
254
not self.to_repository._fetch_uses_deltas))
258
def _fetch_revision_texts(self, revs, pb):
259
# fetch signatures first and then the revision texts
260
# may need to be a InterRevisionStore call here.
261
to_sf = self.to_repository.signatures
262
from_sf = self.from_repository.signatures
263
# A missing signature is just skipped.
264
to_sf.insert_record_stream(filter_absent(from_sf.get_record_stream(
265
[(rev_id,) for rev_id in revs],
266
self.to_repository._fetch_order,
267
not self.to_repository._fetch_uses_deltas)))
268
self._fetch_just_revision_texts(revs)
270
def _fetch_just_revision_texts(self, version_ids):
271
to_rf = self.to_repository.revisions
272
from_rf = self.from_repository.revisions
273
# If a revision has a delta, this is actually expanded inside the
274
# insert_record_stream code now, which is an alternate fix for
276
to_rf.insert_record_stream(from_rf.get_record_stream(
277
[(rev_id,) for rev_id in version_ids],
278
self.to_repository._fetch_order,
279
not self.to_repository._fetch_uses_deltas))
281
def _generate_root_texts(self, revs):
282
"""This will be called by __fetch between fetching weave texts and
283
fetching the inventory weave.
285
Subclasses should override this if they need to generate root texts
286
after fetching weave texts.
291
class Inter1and2Helper(object):
292
"""Helper for operations that convert data from model 1 and 2
294
This is for use by fetchers and converters.
297
def __init__(self, source, target):
300
:param source: The repository data comes from
301
:param target: The repository data goes to
306
def iter_rev_trees(self, revs):
307
"""Iterate through RevisionTrees efficiently.
309
Additionally, the inventory's revision_id is set if unset.
311
Trees are retrieved in batches of 100, and then yielded in the order
314
:param revs: A list of revision ids
316
# In case that revs is not a list.
319
for tree in self.source.revision_trees(revs[:100]):
320
if tree.inventory.revision_id is None:
321
tree.inventory.revision_id = tree.get_revision_id()
325
def _find_root_ids(self, revs, parent_map, graph):
327
planned_versions = {}
328
for tree in self.iter_rev_trees(revs):
329
revision_id = tree.inventory.root.revision
330
root_id = tree.get_root_id()
331
planned_versions.setdefault(root_id, []).append(revision_id)
332
revision_root[revision_id] = root_id
333
# Find out which parents we don't already know root ids for
335
for revision_parents in parent_map.itervalues():
336
parents.update(revision_parents)
337
parents.difference_update(revision_root.keys() + [NULL_REVISION])
338
# Limit to revisions present in the versionedfile
339
parents = graph.get_parent_map(parents).keys()
340
for tree in self.iter_rev_trees(parents):
341
root_id = tree.get_root_id()
342
revision_root[tree.get_revision_id()] = root_id
343
return revision_root, planned_versions
345
def generate_root_texts(self, revs):
346
"""Generate VersionedFiles for all root ids.
348
:param revs: the revisions to include
350
to_texts = self.target.texts
351
graph = self.source.get_graph()
352
parent_map = graph.get_parent_map(revs)
353
rev_order = topo_sort(parent_map)
354
rev_id_to_root_id, root_id_to_rev_ids = self._find_root_ids(
355
revs, parent_map, graph)
356
root_id_order = [(rev_id_to_root_id[rev_id], rev_id) for rev_id in
358
# Guaranteed stable, this groups all the file id operations together
359
# retaining topological order within the revisions of a file id.
360
# File id splits and joins would invalidate this, but they don't exist
361
# yet, and are unlikely to in non-rich-root environments anyway.
362
root_id_order.sort(key=operator.itemgetter(0))
363
# Create a record stream containing the roots to create.
365
for key in root_id_order:
366
root_id, rev_id = key
367
rev_parents = parent_map[rev_id]
368
# We drop revision parents with different file-ids, because
369
# that represents a rename of the root to a different location
370
# - its not actually a parent for us. (We could look for that
371
# file id in the revision tree at considerably more expense,
372
# but for now this is sufficient (and reconcile will catch and
373
# correct this anyway).
374
# When a parent revision is a ghost, we guess that its root id
375
# was unchanged (rather than trimming it from the parent list).
376
parent_keys = tuple((root_id, parent) for parent in rev_parents
377
if parent != NULL_REVISION and
378
rev_id_to_root_id.get(parent, root_id) == root_id)
379
yield FulltextContentFactory(key, parent_keys, None, '')
380
to_texts.insert_record_stream(yield_roots())
382
def regenerate_inventory(self, revs):
383
"""Generate a new inventory versionedfile in target, convertin data.
385
The inventory is retrieved from the source, (deserializing it), and
386
stored in the target (reserializing it in a different format).
387
:param revs: The revisions to include
389
for tree in self.iter_rev_trees(revs):
390
parents = tree.get_parent_ids()
391
self.target.add_inventory(tree.get_revision_id(), tree.inventory,
394
def fetch_revisions(self, revision_ids):
395
# TODO: should this batch them up rather than requesting 10,000
397
for revision in self.source.get_revisions(revision_ids):
398
self.target.add_revision(revision.revision_id, revision)
401
class Model1toKnit2Fetcher(RepoFetcher):
402
"""Fetch from a Model1 repository into a Knit2 repository
404
def __init__(self, to_repository, from_repository, last_revision=None,
405
pb=None, find_ghosts=True):
406
self.helper = Inter1and2Helper(from_repository, to_repository)
407
RepoFetcher.__init__(self, to_repository, from_repository,
408
last_revision, pb, find_ghosts)
410
def _generate_root_texts(self, revs):
411
self.helper.generate_root_texts(revs)
413
def _fetch_inventory_weave(self, revs, pb):
414
self.helper.regenerate_inventory(revs)
416
def _fetch_revision_texts(self, revs, pb):
417
"""Fetch revision object texts"""
421
pb.update('copying revisions', count, total)
423
sig_text = self.from_repository.get_signature_text(rev)
424
self.to_repository.add_signature_text(rev, sig_text)
425
except errors.NoSuchRevision:
428
self._copy_revision(rev)
431
def _copy_revision(self, rev):
432
self.helper.fetch_revisions([rev])
435
class Knit1to2Fetcher(RepoFetcher):
436
"""Fetch from a Knit1 repository into a Knit2 repository"""
438
def __init__(self, to_repository, from_repository, last_revision=None,
439
pb=None, find_ghosts=True):
440
self.helper = Inter1and2Helper(from_repository, to_repository)
441
RepoFetcher.__init__(self, to_repository, from_repository,
442
last_revision, pb, find_ghosts)
444
def _generate_root_texts(self, revs):
445
self.helper.generate_root_texts(revs)
447
def _fetch_inventory_weave(self, revs, pb):
448
self.helper.regenerate_inventory(revs)
450
def _fetch_just_revision_texts(self, version_ids):
451
self.helper.fetch_revisions(version_ids)
added
removed
bzrlib/fetch.py
