1
# Copyright (C) 2005 by Canonical Ltd
1
# Copyright (C) 2005, 2006 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
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):
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.progress import ProgressPhase
39
from bzrlib.revision import NULL_REVISION
40
from bzrlib.symbol_versioning import (deprecated_function,
44
from bzrlib.trace import mutter
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
@deprecated_function(zero_eight)
67
def greedy_fetch(to_branch, from_branch, revision=None, pb=None):
68
"""Legacy API, please see branch.fetch(from_branch, last_revision, pb)."""
27
69
f = Fetcher(to_branch, from_branch, revision, pb)
28
70
return f.count_copied, f.failed_revisions
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
75
class RepoFetcher(object):
76
"""Pull revisions and texts from one repository to another.
79
if set, try to limit to the data this revision references.
82
count_copied -- number of revisions copied
84
This should not be used directory, its essential a object to encapsulate
85
the logic in InterRepository.fetch().
87
def __init__(self, to_repository, from_repository, last_revision=None, pb=None):
89
self.failed_revisions = []
91
if to_repository.control_files._transport.base == from_repository.control_files._transport.base:
92
# check that last_revision is in 'from' and then return a no-operation.
93
if last_revision not in (None, NULL_REVISION):
94
from_repository.get_revision(last_revision)
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
38
self.pb = bzrlib.ui.ui_factory.progress_bar()
101
self.pb = bzrlib.ui.ui_factory.nested_progress_bar()
102
self.nested_pb = self.pb
41
self._scan_histories()
42
self.failed_revisions = []
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]
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
105
self.nested_pb = None
106
self.from_repository.lock_read()
108
self.to_repository.lock_write()
112
if self.nested_pb is not None:
113
self.nested_pb.finished()
114
self.to_repository.unlock()
116
self.from_repository.unlock()
119
"""Primary worker function.
121
This initialises all the needed variables, and then fetches the
122
requested revisions, finally clearing the progress bar.
124
self.to_weaves = self.to_repository.weave_store
125
self.to_control = self.to_repository.control_weaves
126
self.from_weaves = self.from_repository.weave_store
127
self.from_control = self.from_repository.control_weaves
129
self.file_ids_names = {}
130
pp = ProgressPhase('Fetch phase', 4, self.pb)
133
revs = self._revids_to_fetch()
137
self._fetch_weave_texts(revs)
139
self._fetch_inventory_weave(revs)
141
self._fetch_revision_texts(revs)
142
self.count_copied += len(revs)
146
def _revids_to_fetch(self):
147
mutter('fetch up to rev {%s}', self._last_revision)
148
if self._last_revision is NULL_REVISION:
149
# explicit limit of no revisions needed
151
if (self._last_revision is not None and
152
self.to_repository.has_revision(self._last_revision)):
156
return self.to_repository.missing_revision_ids(self.from_repository,
158
except errors.NoSuchRevision:
159
raise InstallFailed([self._last_revision])
161
def _fetch_weave_texts(self, revs):
162
texts_pb = bzrlib.ui.ui_factory.nested_progress_bar()
164
# fileids_altered_by_revision_ids requires reading the inventory
165
# weave, we will need to read the inventory weave again when
166
# all this is done, so enable caching for that specific weave
167
inv_w = self.from_repository.get_inventory_weave()
169
file_ids = self.from_repository.fileids_altered_by_revision_ids(revs)
171
num_file_ids = len(file_ids)
172
for file_id, required_versions in file_ids.items():
173
texts_pb.update("fetch texts", count, num_file_ids)
175
to_weave = self.to_weaves.get_weave_or_empty(file_id,
176
self.to_repository.get_transaction())
177
from_weave = self.from_weaves.get_weave(file_id,
178
self.from_repository.get_transaction())
179
# we fetch all the texts, because texts do
180
# not reference anything, and its cheap enough
181
to_weave.join(from_weave, version_ids=required_versions)
182
# we don't need *all* of this data anymore, but we dont know
183
# what we do. This cache clearing will result in a new read
184
# of the knit data when we do the checkout, but probably we
185
# want to emit the needed data on the fly rather than at the
187
# the from weave should know not to cache data being joined,
188
# but its ok to ask it to clear.
189
from_weave.clear_cache()
190
to_weave.clear_cache()
194
def _fetch_inventory_weave(self, revs):
195
pb = bzrlib.ui.ui_factory.nested_progress_bar()
197
pb.update("fetch inventory", 0, 2)
198
to_weave = self.to_control.get_weave('inventory',
199
self.to_repository.get_transaction())
201
child_pb = bzrlib.ui.ui_factory.nested_progress_bar()
203
# just merge, this is optimisable and its means we don't
204
# copy unreferenced data such as not-needed inventories.
205
pb.update("fetch inventory", 1, 3)
206
from_weave = self.from_repository.get_inventory_weave()
207
pb.update("fetch inventory", 2, 3)
208
# we fetch only the referenced inventories because we do not
209
# know for unselected inventories whether all their required
210
# texts are present in the other repository - it could be
212
to_weave.join(from_weave, pb=child_pb, msg='merge inventory',
214
from_weave.clear_cache()
221
class GenericRepoFetcher(RepoFetcher):
222
"""This is a generic repo to repo fetcher.
224
This makes minimal assumptions about repo layout and contents.
225
It triggers a reconciliation after fetching to ensure integrity.
228
def _fetch_revision_texts(self, revs):
229
"""Fetch revision object texts"""
230
rev_pb = bzrlib.ui.ui_factory.nested_progress_bar()
232
to_txn = self.to_transaction = self.to_repository.get_transaction()
235
to_store = self.to_repository._revision_store
237
pb = bzrlib.ui.ui_factory.nested_progress_bar()
239
pb.update('copying revisions', count, total)
241
sig_text = self.from_repository.get_signature_text(rev)
242
to_store.add_revision_signature_text(rev, sig_text, to_txn)
243
except errors.NoSuchRevision:
246
to_store.add_revision(self.from_repository.get_revision(rev),
251
# fixup inventory if needed:
252
# this is expensive because we have no inverse index to current ghosts.
253
# but on local disk its a few seconds and sftp push is already insane.
255
# FIXME: repository should inform if this is needed.
256
self.to_repository.reconcile()
261
class KnitRepoFetcher(RepoFetcher):
262
"""This is a knit format repository specific fetcher.
264
This differs from the GenericRepoFetcher by not doing a
265
reconciliation after copying, and using knit joining to
269
def _fetch_revision_texts(self, revs):
270
# may need to be a InterRevisionStore call here.
271
from_transaction = self.from_repository.get_transaction()
272
to_transaction = self.to_repository.get_transaction()
273
to_sf = self.to_repository._revision_store.get_signature_file(
275
from_sf = self.from_repository._revision_store.get_signature_file(
277
to_sf.join(from_sf, version_ids=revs, ignore_missing=True)
278
to_rf = self.to_repository._revision_store.get_revision_file(
280
from_rf = self.from_repository._revision_store.get_revision_file(
282
to_rf.join(from_rf, version_ids=revs)
285
class Inter1and2Helper(object):
286
"""Helper for operations that convert data from model 1 and 2
288
This is for use by fetchers and converters.
291
def __init__(self, source, target):
294
:param source: The repository data comes from
295
:param target: The repository data goes to
300
def iter_rev_trees(self, revs):
301
"""Iterate through RevisionTrees efficiently.
303
Additionally, the inventory's revision_id is set if unset.
305
Trees are retrieved in batches of 100, and then yielded in the order
308
:param revs: A list of revision ids
311
for tree in self.source.revision_trees(revs[:100]):
312
if tree.inventory.revision_id is None:
313
tree.inventory.revision_id = tree.get_revision_id()
317
def generate_root_texts(self, revs):
318
"""Generate VersionedFiles for all root ids.
320
:param revs: the revisions to include
322
inventory_weave = self.source.get_inventory_weave()
325
to_store = self.target.weave_store
326
for tree in self.iter_rev_trees(revs):
327
revision_id = tree.inventory.root.revision
328
root_id = tree.inventory.root.file_id
329
parents = inventory_weave.get_parents(revision_id)
330
if root_id not in versionedfile:
331
versionedfile[root_id] = to_store.get_weave_or_empty(root_id,
332
self.target.get_transaction())
333
parent_texts[root_id] = versionedfile[root_id].add_lines(
334
revision_id, parents, [], parent_texts)
336
def regenerate_inventory(self, revs):
337
"""Generate a new inventory versionedfile in target, convertin data.
339
The inventory is retrieved from the source, (deserializing it), and
340
stored in the target (reserializing it in a different format).
341
:param revs: The revisions to include
343
inventory_weave = self.source.get_inventory_weave()
344
for tree in self.iter_rev_trees(revs):
345
parents = inventory_weave.get_parents(tree.get_revision_id())
346
self.target.add_inventory(tree.get_revision_id(), tree.inventory,
350
class Model1toKnit2Fetcher(GenericRepoFetcher):
351
"""Fetch from a Model1 repository into a Knit2 repository
353
def __init__(self, to_repository, from_repository, last_revision=None,
355
self.helper = Inter1and2Helper(from_repository, to_repository)
356
GenericRepoFetcher.__init__(self, to_repository, from_repository,
359
def _fetch_weave_texts(self, revs):
360
GenericRepoFetcher._fetch_weave_texts(self, revs)
361
# Now generate a weave for the tree root
362
self.helper.generate_root_texts(revs)
364
def _fetch_inventory_weave(self, revs):
365
self.helper.regenerate_inventory(revs)
368
class Knit1to2Fetcher(KnitRepoFetcher):
369
"""Fetch from a Knit1 repository into a Knit2 repository"""
371
def __init__(self, to_repository, from_repository, last_revision=None,
373
self.helper = Inter1and2Helper(from_repository, to_repository)
374
KnitRepoFetcher.__init__(self, to_repository, from_repository,
377
def _fetch_weave_texts(self, revs):
378
KnitRepoFetcher._fetch_weave_texts(self, revs)
379
# Now generate a weave for the tree root
380
self.helper.generate_root_texts(revs)
382
def _fetch_inventory_weave(self, revs):
383
self.helper.regenerate_inventory(revs)
386
class Fetcher(object):
387
"""Backwards compatibility glue for branch.fetch()."""
389
@deprecated_method(zero_eight)
390
def __init__(self, to_branch, from_branch, last_revision=None, pb=None):
391
"""Please see branch.fetch()."""
392
to_branch.fetch(from_branch, last_revision, pb)
added
removed
bzrlib/fetch.py
