67
68
return f.count_copied, f.failed_revisions
71
class RepoFetcher(object):
72
"""Pull revisions and texts from one repository to another.
75
if set, try to limit to the data this revision references.
78
count_copied -- number of revisions copied
79
count_weaves -- number of file weaves copied
81
def __init__(self, to_repository, from_repository, last_revision=None, pb=None):
82
if to_repository.bzrdir.transport.base == from_repository.bzrdir.transport.base:
83
# check that last_revision is in 'from' and then return a no-operation.
84
if last_revision not in (None, NULL_REVISION):
85
from_repository.get_revision(last_revision)
87
self.to_repository = to_repository
88
self.from_repository = from_repository
89
# must not mutate self._last_revision as its potentially a shared instance
90
self._last_revision = last_revision
92
self.pb = bzrlib.ui.ui_factory.progress_bar()
95
self.from_repository.lock_read()
97
self.to_repository.lock_write()
101
self.to_repository.unlock()
103
self.from_repository.unlock()
106
"""Primary worker function.
108
This initialises all the needed variables, and then fetches the
109
requested revisions, finally clearing the progress bar.
111
self.to_weaves = self.to_repository.weave_store
112
self.to_control = self.to_repository.control_weaves
113
self.from_weaves = self.from_repository.weave_store
114
self.from_control = self.from_repository.control_weaves
115
self.failed_revisions = []
116
self.count_copied = 0
118
self.count_weaves = 0
119
self.copied_file_ids = set()
120
self.file_ids_names = {}
122
revs = self._revids_to_fetch()
125
self._fetch_weave_texts(revs)
126
self._fetch_inventory_weave(revs)
127
self._fetch_revision_texts(revs)
128
self.count_copied += len(revs)
132
def _revids_to_fetch(self):
133
mutter('fetch up to rev {%s}', self._last_revision)
134
if self._last_revision is NULL_REVISION:
135
# explicit limit of no revisions needed
137
if (self._last_revision != None and
138
self.to_repository.has_revision(self._last_revision)):
142
return self.to_repository.missing_revision_ids(self.from_repository,
144
except errors.NoSuchRevision:
145
raise InstallFailed([self._last_revision])
147
def _fetch_revision_texts(self, revs):
148
self.to_repository.revision_store.copy_multi(
149
self.from_repository.revision_store,
153
def _fetch_weave_texts(self, revs):
154
file_ids = self.from_repository.fileid_involved_by_set(revs)
156
num_file_ids = len(file_ids)
157
for file_id in file_ids:
158
self.pb.update("merge weaves", count, num_file_ids)
160
to_weave = self.to_weaves.get_weave_or_empty(file_id,
161
self.to_repository.get_transaction())
162
from_weave = self.from_weaves.get_weave(file_id,
163
self.from_repository.get_transaction())
165
if to_weave.numversions() > 0:
166
# destination has contents, must merge
168
to_weave.join(from_weave)
169
except errors.WeaveParentMismatch:
170
to_weave.reweave(from_weave)
172
# destination is empty, just replace it
173
to_weave = from_weave.copy()
175
self.to_weaves.put_weave(file_id, to_weave,
176
self.to_repository.get_transaction())
179
def _fetch_inventory_weave(self, revs):
180
self.pb.update("inventory fetch", 0, 2)
181
from_weave = self.from_repository.get_inventory_weave()
182
to_weave = self.to_repository.get_inventory_weave()
183
self.pb.update("inventory fetch", 1, 2)
184
to_weave = self.to_control.get_weave('inventory',
185
self.to_repository.get_transaction())
186
self.pb.update("inventory fetch", 2, 2)
188
if to_weave.numversions() > 0:
189
# destination has contents, must merge
191
to_weave.join(from_weave, pb=self.pb, msg='merge inventory')
192
except errors.WeaveParentMismatch:
193
to_weave.reweave(from_weave, pb=self.pb, msg='reweave inventory')
195
# destination is empty, just replace it
196
to_weave = from_weave.copy()
198
self.to_control.put_weave('inventory', to_weave,
199
self.to_repository.get_transaction())
71
204
class Fetcher(object):
72
205
"""Pull revisions and texts from one branch to another.
88
221
count_weaves -- number of file weaves copied
90
223
def __init__(self, to_branch, from_branch, last_revision=None, pb=None):
91
if to_branch == from_branch:
92
raise Exception("can't fetch from a branch to itself")
224
if to_branch.base == from_branch.base:
225
raise Exception("can't fetch from a branch to itself %s, %s" %
226
(from_branch.base, to_branch.base))
93
228
self.to_branch = to_branch
94
self.to_weaves = to_branch.weave_store
95
self.to_control = to_branch.control_weaves
96
229
self.from_branch = from_branch
97
self.from_weaves = from_branch.weave_store
98
self.from_control = from_branch.control_weaves
99
self.failed_revisions = []
100
self.count_copied = 0
102
self.count_weaves = 0
103
self.copied_file_ids = set()
104
self.file_ids_names = {}
230
self._last_revision = last_revision
106
232
self.pb = bzrlib.ui.ui_factory.progress_bar()
109
235
self.from_branch.lock_read()
111
self._fetch_revisions(last_revision)
237
self.to_branch.lock_write()
241
self.to_branch.unlock()
113
243
self.from_branch.unlock()
116
def _fetch_revisions(self, last_revision):
117
self.last_revision = self._find_last_revision(last_revision)
118
mutter('fetch up to rev {%s}', self.last_revision)
119
if (self.last_revision is not None and
120
self.to_branch.has_revision(self.last_revision)):
123
revs_to_fetch = self._compare_ancestries()
125
raise InstallFailed([self.last_revision])
126
self._copy_revisions(revs_to_fetch)
127
self.new_ancestry = revs_to_fetch
129
def _find_last_revision(self, last_revision):
246
self._find_last_revision()
247
repo_fetcher = RepoFetcher(to_repository=self.to_branch.repository,
248
from_repository=self.from_branch.repository,
250
last_revision=self._last_revision)
251
self.failed_revisions = repo_fetcher.failed_revisions
252
self.count_copied = repo_fetcher.count_copied
253
self.count_total = repo_fetcher.count_total
254
self.count_weaves = repo_fetcher.count_weaves
255
self.copied_file_ids = repo_fetcher.copied_file_ids
257
def _find_last_revision(self):
130
258
"""Find the limiting source revision.
132
260
Every ancestor of that revision will be merged across.
134
262
Returns the revision_id, or returns None if there's no history
135
263
in the source branch."""
264
if self._last_revision:
138
266
self.pb.update('get source history')
139
267
from_history = self.from_branch.revision_history()
140
268
self.pb.update('get destination history')
142
return from_history[-1]
144
return None # no history in the source branch
147
def _compare_ancestries(self):
148
"""Get a list of revisions that must be copied.
150
That is, every revision that's in the ancestry of the source
151
branch and not in the destination branch."""
152
self.pb.update('get source ancestry')
153
self.from_ancestry = self.from_branch.get_ancestry(self.last_revision)
155
dest_last_rev = self.to_branch.last_revision()
156
self.pb.update('get destination ancestry')
158
dest_ancestry = self.to_branch.get_ancestry(dest_last_rev)
161
ss = set(dest_ancestry)
163
for rev_id in self.from_ancestry:
165
to_fetch.append(rev_id)
166
mutter('need to get revision {%s}', rev_id)
167
mutter('need to get %d revisions in total', len(to_fetch))
168
self.count_total = len(to_fetch)
171
def _copy_revisions(self, revs_to_fetch):
173
for rev_id in revs_to_fetch:
177
if self.to_branch.has_revision(rev_id):
179
self.pb.update('copy revision', i, self.count_total)
180
self._copy_one_revision(rev_id)
181
self.count_copied += 1
184
def _copy_one_revision(self, rev_id):
185
"""Copy revision and everything referenced by it."""
186
mutter('copying revision {%s}', rev_id)
187
rev_xml = self.from_branch.get_revision_xml(rev_id)
188
inv_xml = self.from_branch.get_inventory_xml(rev_id)
189
rev = serializer_v5.read_revision_from_string(rev_xml)
190
inv = serializer_v5.read_inventory_from_string(inv_xml)
191
assert rev.revision_id == rev_id
192
assert rev.inventory_sha1 == sha_string(inv_xml)
193
mutter(' commiter %s, %d parents',
196
self._copy_new_texts(rev_id, inv)
197
parents = rev.parent_ids
198
new_parents = copy(parents)
199
for parent in parents:
200
if not self.to_branch.has_revision(parent):
201
new_parents.pop(new_parents.index(parent))
202
self._copy_inventory(rev_id, inv_xml, new_parents)
203
self.to_branch.revision_store.add(StringIO(rev_xml), rev_id)
204
mutter('copied revision %s', rev_id)
206
def _copy_inventory(self, rev_id, inv_xml, parent_ids):
207
self.to_control.add_text('inventory', rev_id,
208
split_lines(inv_xml), parent_ids,
209
self.to_branch.get_transaction())
211
def _copy_new_texts(self, rev_id, inv):
212
"""Copy any new texts occuring in this revision."""
213
# TODO: Rather than writing out weaves every time, hold them
214
# in memory until everything's done? But this way is nicer
215
# if it's interrupted.
216
for path, ie in inv.iter_entries():
217
self._copy_one_weave(rev_id, ie.file_id, ie.revision)
219
def _copy_one_weave(self, rev_id, file_id, text_revision):
220
"""Copy one file weave, esuring the result contains text_revision."""
221
# check if the revision is already there
222
if file_id in self.file_ids_names.keys( ) and \
223
text_revision in self.file_ids_names[file_id]:
225
to_weave = self.to_weaves.get_weave_or_empty(file_id,
226
self.to_branch.get_transaction())
227
if not file_id in self.file_ids_names.keys( ):
228
self.file_ids_names[file_id] = to_weave.names( )
229
if text_revision in to_weave:
231
from_weave = self.from_weaves.get_weave(file_id,
232
self.from_branch.get_transaction())
233
if text_revision not in from_weave:
234
raise MissingText(self.from_branch, text_revision, file_id)
235
mutter('copy file {%s} modified in {%s}', file_id, rev_id)
237
if to_weave.numversions() > 0:
238
# destination has contents, must merge
240
to_weave.join(from_weave)
241
except errors.WeaveParentMismatch:
242
to_weave.reweave(from_weave)
244
# destination is empty, just replace it
245
to_weave = from_weave.copy( )
246
self.to_weaves.put_weave(file_id, to_weave,
247
self.to_branch.get_transaction())
248
self.count_weaves += 1
249
self.copied_file_ids.add(file_id)
250
self.file_ids_names[file_id] = to_weave.names()
251
mutter('copied file {%s}', file_id)
270
self._last_revision = from_history[-1]
272
# no history in the source branch
273
self._last_revision = NULL_REVISION
added
removed
bzrlib/fetch.py
