1
# Copyright (C) 2005, 2006 Canonical Ltd
1
# Copyright (C) 2005 by Canonical Ltd
3
3
# This program is free software; you can redistribute it and/or modify
4
4
# it under the terms of the GNU General Public License as published by
5
5
# the Free Software Foundation; either version 2 of the License, or
6
6
# (at your option) any later version.
8
8
# This program is distributed in the hope that it will be useful,
9
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
11
# GNU General Public License for more details.
13
13
# You should have received a copy of the GNU General Public License
14
14
# along with this program; if not, write to the Free Software
15
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 finally actually join
35
import bzrlib.errors as errors
36
from bzrlib.errors import (InstallFailed,
38
from bzrlib.trace import mutter
39
from bzrlib.progress import ProgressPhase
40
from bzrlib.revision import NULL_REVISION
41
from bzrlib.symbol_versioning import (deprecated_function,
47
# TODO: Avoid repeatedly opening weaves so many times.
49
# XXX: This doesn't handle ghost (not present in branch) revisions at
50
# all yet. I'm not sure they really should be supported.
52
# NOTE: This doesn't copy revisions which may be present but not
53
# merged into the last revision. I'm not sure we want to do that.
55
# - get a list of revisions that need to be pulled in
56
# - for each one, pull in that revision file
57
# and get the inventory, and store the inventory with right
59
# - and get the ancestry, and store that with right parents too
60
# - and keep a note of all file ids and version seen
61
# - then go through all files; for each one get the weave,
62
# and add in all file versions
65
@deprecated_function(zero_eight)
66
def greedy_fetch(to_branch, from_branch, revision=None, pb=None):
67
"""Legacy API, please see branch.fetch(from_branch, last_revision, pb)."""
21
from bzrlib.trace import mutter, note
22
from bzrlib.branch import Branch
23
from bzrlib.progress import ProgressBar
26
def greedy_fetch(to_branch, from_branch, revision, pb):
68
27
f = Fetcher(to_branch, from_branch, revision, pb)
69
28
return f.count_copied, f.failed_revisions
74
class RepoFetcher(object):
75
"""Pull revisions and texts from one repository to another.
78
if set, try to limit to the data this revision references.
81
count_copied -- number of revisions copied
83
This should not be used directory, its essential a object to encapsulate
84
the logic in InterRepository.fetch().
86
def __init__(self, to_repository, from_repository, last_revision=None, pb=None):
31
class Fetcher(object):
32
"""Pull history from one branch to another."""
33
def __init__(self, to_branch, from_branch, revision_limit=None, pb=None):
34
self.to_branch = to_branch
35
self.from_branch = from_branch
36
self.revision_limit = revision_limit
38
self.pb = bzrlib.ui.ui_factory.progress_bar()
41
self._scan_histories()
88
42
self.failed_revisions = []
89
43
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)
95
self.to_repository = to_repository
96
self.from_repository = from_repository
97
# must not mutate self._last_revision as its potentially a shared instance
98
self._last_revision = last_revision
100
self.pb = bzrlib.ui.ui_factory.nested_progress_bar()
101
self.nested_pb = self.pb
47
def _scan_histories(self):
48
self.from_history = from_branch.revision_history()
49
self.required_revisions = set(from_history)
50
self.to_history = to_branch.revision_history()
51
if self.revision_limit:
52
raise NotImplementedError('sorry, revision_limit not handled yet')
53
self.need_revisions = []
54
for rev_id in self.from_history:
55
if not has_revision(self.to_branch):
56
self.need_revisions.append(rev_id)
57
mutter('need to get revision {%s}', rev_id)
61
while self.need_revisions:
62
rev_id = self.need_revisions.pop()
63
mutter('try to get revision {%s}', rev_id)
69
def has_revision(branch, revision_id):
71
branch.get_revision_xml(revision_id)
73
except bzrlib.errors.NoSuchRevision:
77
def old_greedy_fetch(to_branch, from_branch, revision=None, pb=None):
78
"""Copy all history from one branch to another.
81
If set, copy only up to this point in the source branch.
83
@returns: number copied, missing ids
85
from_history = from_branch.revision_history()
86
required_revisions = set(from_history)
88
if revision is not None:
89
required_revisions.add(revision)
91
rev_index = from_history.index(revision)
94
if rev_index is not None:
95
from_history = from_history[:rev_index + 1]
104
self.nested_pb = None
105
self.from_repository.lock_read()
107
self.to_repository.lock_write()
111
if self.nested_pb is not None:
112
self.nested_pb.finished()
113
self.to_repository.unlock()
115
self.from_repository.unlock()
118
"""Primary worker function.
120
This initialises all the needed variables, and then fetches the
121
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
128
self.file_ids_names = {}
129
pp = ProgressPhase('Fetch phase', 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)
145
def _revids_to_fetch(self):
146
mutter('fetch up to rev {%s}', self._last_revision)
147
if self._last_revision is NULL_REVISION:
148
# explicit limit of no revisions needed
150
if (self._last_revision is not None and
151
self.to_repository.has_revision(self._last_revision)):
155
return self.to_repository.missing_revision_ids(self.from_repository,
157
except errors.NoSuchRevision:
158
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):
269
# 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)
284
class Inter1and2Helper(object):
285
"""Helper for operations that convert data from model 1 and 2
287
This is for use by fetchers and converters.
290
def __init__(self, source, target):
293
:param source: The repository data comes from
294
:param target: The repository data goes to
299
def iter_rev_trees(self, revs):
300
"""Iterate through RevisionTrees efficiently.
302
Additionally, the inventory's revision_id is set if unset.
304
Trees are retrieved in batches of 100, and then yielded in the order
307
:param revs: A list of revision ids
310
for tree in self.source.revision_trees(revs[:100]):
311
if tree.inventory.revision_id is None:
312
tree.inventory.revision_id = tree.get_revision_id()
316
def generate_root_texts(self, revs):
317
"""Generate VersionedFiles for all root ids.
319
: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)
335
def regenerate_inventory(self, revs):
336
"""Generate a new inventory versionedfile in target, convertin data.
338
The inventory is retrieved from the source, (deserializing it), and
339
stored in the target (reserializing it in a different format).
340
:param revs: The revisions to include
342
inventory_weave = self.source.get_inventory_weave()
343
for tree in self.iter_rev_trees(revs):
344
parents = inventory_weave.get_parents(tree.get_revision_id())
345
self.target.add_inventory(tree.get_revision_id(), tree.inventory,
349
class Model1toKnit2Fetcher(GenericRepoFetcher):
350
"""Fetch from a Model1 repository into a Knit2 repository
352
def __init__(self, to_repository, from_repository, last_revision=None,
354
self.helper = Inter1and2Helper(from_repository, to_repository)
355
GenericRepoFetcher.__init__(self, to_repository, from_repository,
358
def _fetch_weave_texts(self, revs):
359
GenericRepoFetcher._fetch_weave_texts(self, revs)
360
# Now generate a weave for the tree root
361
self.helper.generate_root_texts(revs)
363
def _fetch_inventory_weave(self, revs):
364
self.helper.regenerate_inventory(revs)
367
class Knit1to2Fetcher(KnitRepoFetcher):
368
"""Fetch from a Knit1 repository into a Knit2 repository"""
370
def __init__(self, to_repository, from_repository, last_revision=None,
372
self.helper = Inter1and2Helper(from_repository, to_repository)
373
KnitRepoFetcher.__init__(self, to_repository, from_repository,
376
def _fetch_weave_texts(self, revs):
377
KnitRepoFetcher._fetch_weave_texts(self, revs)
378
# Now generate a weave for the tree root
379
self.helper.generate_root_texts(revs)
381
def _fetch_inventory_weave(self, revs):
382
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)
97
from_history = [revision]
98
to_history = to_branch.revision_history()
100
for rev_id in from_history:
101
if not has_revision(to_branch, rev_id):
102
missing.append(rev_id)
104
# recurse down through the revision graph, looking for things that
107
while len(missing) > 0:
108
installed, failed = to_branch.install_revisions(from_branch,
109
revision_ids=missing,
112
required_failed = failed.intersection(required_revisions)
113
if len(required_failed) > 0:
114
raise bzrlib.errors.InstallFailed(required_failed)
115
for rev_id in failed:
116
note("Failed to install %s" % rev_id)
117
all_failed.update(failed)
119
for rev_id in missing:
121
revision = from_branch.get_revision(rev_id)
122
except bzrlib.errors.NoSuchRevision:
123
if revision in from_history:
127
for parent in [p.revision_id for p in revision.parents]:
128
if not has_revision(to_branch, parent):
129
new_missing.add(parent)
130
missing = new_missing
131
return count, all_failed
134
def old_install_revisions(branch, other, revision_ids, pb):
135
"""Copy revisions from other branch into branch.
137
This is a lower-level function used by a pull or a merge. It
138
incorporates some history from one branch into another, but
139
does not update the revision history or operate on the working
143
Sequence of revisions to copy.
146
Progress bar for copying.
149
if hasattr(other.revision_store, "prefetch"):
150
other.revision_store.prefetch(revision_ids)
151
if hasattr(other.inventory_store, "prefetch"):
152
other.inventory_store.prefetch(revision_ids)
155
pb = bzrlib.ui.ui_factory.progress_bar()
162
for i, rev_id in enumerate(revision_ids):
163
pb.update('fetching revision', i+1, len(revision_ids))
165
rev = other.get_revision(rev_id)
166
except bzrlib.errors.NoSuchRevision:
170
revisions.append(rev)
171
inv = other.get_inventory(rev_id)
172
for key, entry in inv.iter_entries():
173
if entry.text_id is None:
175
if entry.text_id not in branch.text_store:
176
needed_texts.add(entry.text_id)
180
count, cp_fail = branch.text_store.copy_multi(other.text_store,
182
count, cp_fail = branch.inventory_store.copy_multi(other.inventory_store,
184
count, cp_fail = branch.revision_store.copy_multi(other.revision_store,
187
assert len(cp_fail) == 0
188
return count, failures
added
removed
bzrlib/fetch.py
