49
36
def do(self, path, *args):
50
37
"""Execute a repository request.
52
All Repository requests take a path to the repository as their first
53
argument. The repository must be at the exact path given by the
54
client - no searching is done.
39
The repository must be at the exact path - no searching is done.
56
41
The actual logic is delegated to self.do_repository_request.
58
:param client_path: The path for the repository as received from the
60
:return: A SmartServerResponse from self.do_repository_request().
43
:param path: The path for the repository.
44
:return: A smart server from self.do_repository_request().
62
transport = self.transport_from_client_path(path)
46
transport = self._backing_transport.clone(path)
63
47
bzrdir = BzrDir.open_from_transport(transport)
64
# Save the repository for use with do_body.
65
self._repository = bzrdir.open_repository()
66
return self.do_repository_request(self._repository, *args)
68
def do_repository_request(self, repository, *args):
69
"""Override to provide an implementation for a verb."""
70
# No-op for verbs that take bodies (None as a result indicates a body
74
def recreate_search(self, repository, search_bytes):
75
lines = search_bytes.split('\n')
76
if lines[0] == 'ancestry-of':
78
search_result = graph.PendingAncestryResult(heads, repository)
79
return search_result, None
80
elif lines[0] == 'search':
81
return self.recreate_search_from_recipe(repository, lines[1:])
83
return (None, FailedSmartServerResponse(('BadSearch',)))
85
def recreate_search_from_recipe(self, repository, lines):
86
start_keys = set(lines[0].split(' '))
87
exclude_keys = set(lines[1].split(' '))
88
revision_count = int(lines[2])
89
repository.lock_read()
91
search = repository.get_graph()._make_breadth_first_searcher(
95
next_revs = search.next()
98
search.stop_searching_any(exclude_keys.intersection(next_revs))
99
search_result = search.get_result()
100
if search_result.get_recipe()[3] != revision_count:
101
# we got back a different amount of data than expected, this
102
# gets reported as NoSuchRevision, because less revisions
103
# indicates missing revisions, and more should never happen as
104
# the excludes list considers ghosts and ensures that ghost
105
# filling races are not a problem.
106
return (None, FailedSmartServerResponse(('NoSuchRevision',)))
107
return (search_result, None)
112
class SmartServerRepositoryReadLocked(SmartServerRepositoryRequest):
113
"""Calls self.do_readlocked_repository_request."""
115
def do_repository_request(self, repository, *args):
116
"""Read lock a repository for do_readlocked_repository_request."""
117
repository.lock_read()
119
return self.do_readlocked_repository_request(repository, *args)
124
class SmartServerRepositoryGetParentMap(SmartServerRepositoryRequest):
125
"""Bzr 1.2+ - get parent data for revisions during a graph search."""
127
no_extra_results = False
129
def do_repository_request(self, repository, *revision_ids):
130
"""Get parent details for some revisions.
132
All the parents for revision_ids are returned. Additionally up to 64KB
133
of additional parent data found by performing a breadth first search
134
from revision_ids is returned. The verb takes a body containing the
135
current search state, see do_body for details.
137
If 'include-missing:' is in revision_ids, ghosts encountered in the
138
graph traversal for getting parent data are included in the result with
139
a prefix of 'missing:'.
141
:param repository: The repository to query in.
142
:param revision_ids: The utf8 encoded revision_id to answer for.
144
self._revision_ids = revision_ids
145
return None # Signal that we want a body.
147
def do_body(self, body_bytes):
148
"""Process the current search state and perform the parent lookup.
150
:return: A smart server response where the body contains an utf8
151
encoded flattened list of the parents of the revisions (the same
152
format as Repository.get_revision_graph) which has been bz2
155
repository = self._repository
156
repository.lock_read()
158
return self._do_repository_request(body_bytes)
162
def _do_repository_request(self, body_bytes):
163
repository = self._repository
164
revision_ids = set(self._revision_ids)
165
include_missing = 'include-missing:' in revision_ids
167
revision_ids.remove('include-missing:')
168
body_lines = body_bytes.split('\n')
169
search_result, error = self.recreate_search_from_recipe(
170
repository, body_lines)
171
if error is not None:
173
# TODO might be nice to start up the search again; but thats not
174
# written or tested yet.
175
client_seen_revs = set(search_result.get_keys())
176
# Always include the requested ids.
177
client_seen_revs.difference_update(revision_ids)
179
repo_graph = repository.get_graph()
183
next_revs = revision_ids
184
first_loop_done = False
186
queried_revs.update(next_revs)
187
parent_map = repo_graph.get_parent_map(next_revs)
188
current_revs = next_revs
190
for revision_id in current_revs:
192
parents = parent_map.get(revision_id)
193
if parents is not None:
194
# adjust for the wire
195
if parents == (_mod_revision.NULL_REVISION,):
197
# prepare the next query
198
next_revs.update(parents)
199
encoded_id = revision_id
202
encoded_id = "missing:" + revision_id
204
if (revision_id not in client_seen_revs and
205
(not missing_rev or include_missing)):
206
# Client does not have this revision, give it to it.
207
# add parents to the result
208
result[encoded_id] = parents
209
# Approximate the serialized cost of this revision_id.
210
size_so_far += 2 + len(encoded_id) + sum(map(len, parents))
211
# get all the directly asked for parents, and then flesh out to
212
# 64K (compressed) or so. We do one level of depth at a time to
213
# stay in sync with the client. The 250000 magic number is
214
# estimated compression ratio taken from bzr.dev itself.
215
if self.no_extra_results or (
216
first_loop_done and size_so_far > 250000):
219
# don't query things we've already queried
220
next_revs.difference_update(queried_revs)
221
first_loop_done = True
223
# sorting trivially puts lexographically similar revision ids together.
225
for revision, parents in sorted(result.items()):
226
lines.append(' '.join((revision, ) + tuple(parents)))
228
return SuccessfulSmartServerResponse(
229
('ok', ), bz2.compress('\n'.join(lines)))
232
class SmartServerRepositoryGetRevisionGraph(SmartServerRepositoryReadLocked):
234
def do_readlocked_repository_request(self, repository, revision_id):
48
repository = bzrdir.open_repository()
49
return self.do_repository_request(repository, *args)
52
class SmartServerRepositoryGetRevisionGraph(SmartServerRepositoryRequest):
54
def do_repository_request(self, repository, revision_id):
235
55
"""Return the result of repository.get_revision_graph(revision_id).
237
Deprecated as of bzr 1.4, but supported for older clients.
239
57
:param repository: The repository to query in.
240
58
:param revision_id: The utf8 encoded revision_id to get a graph from.
241
59
:return: A smart server response where the body contains an utf8
352
163
return FailedSmartServerResponse(('LockContention',))
353
164
except errors.UnlockableTransport:
354
165
return FailedSmartServerResponse(('UnlockableTransport',))
355
except errors.LockFailed, e:
356
return FailedSmartServerResponse(('LockFailed',
357
str(e.lock), str(e.why)))
358
if token is not None:
359
repository.leave_lock_in_place()
166
repository.leave_lock_in_place()
360
167
repository.unlock()
361
168
if token is None:
363
170
return SuccessfulSmartServerResponse(('ok', token))
366
class SmartServerRepositoryGetStream(SmartServerRepositoryRequest):
368
def do_repository_request(self, repository, to_network_name):
369
"""Get a stream for inserting into a to_format repository.
371
:param repository: The repository to stream from.
372
:param to_network_name: The network name of the format of the target
375
self._to_format = network_format_registry.get(to_network_name)
376
return None # Signal that we want a body.
378
def do_body(self, body_bytes):
379
repository = self._repository
380
repository.lock_read()
382
search_result, error = self.recreate_search(repository, body_bytes)
383
if error is not None:
386
source = repository._get_source(self._to_format)
387
stream = source.get_stream(search_result)
389
exc_info = sys.exc_info()
391
# On non-error, unlocking is done by the body stream handler.
394
raise exc_info[0], exc_info[1], exc_info[2]
395
return SuccessfulSmartServerResponse(('ok',),
396
body_stream=self.body_stream(stream, repository))
398
def body_stream(self, stream, repository):
399
byte_stream = _stream_to_byte_stream(stream, repository._format)
401
for bytes in byte_stream:
403
except errors.RevisionNotPresent, e:
404
# This shouldn't be able to happen, but as we don't buffer
405
# everything it can in theory happen.
407
yield FailedSmartServerResponse(('NoSuchRevision', e.revision_id))
412
def _stream_to_byte_stream(stream, src_format):
413
"""Convert a record stream to a self delimited byte stream."""
414
pack_writer = pack.ContainerSerialiser()
415
yield pack_writer.begin()
416
yield pack_writer.bytes_record(src_format.network_name(), '')
417
for substream_type, substream in stream:
418
for record in substream:
419
if record.storage_kind in ('chunked', 'fulltext'):
420
serialised = record_to_fulltext_bytes(record)
422
serialised = record.get_bytes_as(record.storage_kind)
424
# Some streams embed the whole stream into the wire
425
# representation of the first record, which means that
426
# later records have no wire representation: we skip them.
427
yield pack_writer.bytes_record(serialised, [(substream_type,)])
428
yield pack_writer.end()
431
def _byte_stream_to_stream(byte_stream):
432
"""Convert a byte stream into a format and a stream.
434
:param byte_stream: A bytes iterator, as output by _stream_to_byte_stream.
435
:return: (RepositoryFormat, stream_generator)
437
stream_decoder = pack.ContainerPushParser()
439
"""Closure to return the substreams."""
440
# May have fully parsed records already.
441
for record in stream_decoder.read_pending_records():
442
record_names, record_bytes = record
443
record_name, = record_names
444
substream_type = record_name[0]
445
substream = NetworkRecordStream([record_bytes])
446
yield substream_type, substream.read()
447
for bytes in byte_stream:
448
stream_decoder.accept_bytes(bytes)
449
for record in stream_decoder.read_pending_records():
450
record_names, record_bytes = record
451
record_name, = record_names
452
substream_type = record_name[0]
453
substream = NetworkRecordStream([record_bytes])
454
yield substream_type, substream.read()
455
for bytes in byte_stream:
456
stream_decoder.accept_bytes(bytes)
457
for record in stream_decoder.read_pending_records(max=1):
458
record_names, src_format_name = record
459
src_format = network_format_registry.get(src_format_name)
460
return src_format, record_stream()
463
173
class SmartServerRepositoryUnlock(SmartServerRepositoryRequest):
465
175
def do_repository_request(self, repository, token):
534
235
dirname = dirname.encode(sys.getfilesystemencoding())
535
236
# python's tarball module includes the whole path by default so
537
if not dirname.endswith('.bzr'):
538
raise ValueError(dirname)
238
assert dirname.endswith('.bzr')
539
239
tarball.add(dirname, '.bzr') # recursive by default
544
class SmartServerRepositoryInsertStreamLocked(SmartServerRepositoryRequest):
545
"""Insert a record stream from a RemoteSink into a repository.
547
This gets bytes pushed to it by the network infrastructure and turns that
548
into a bytes iterator using a thread. That is then processed by
549
_byte_stream_to_stream.
554
def do_repository_request(self, repository, resume_tokens, lock_token):
555
"""StreamSink.insert_stream for a remote repository."""
556
repository.lock_write(token=lock_token)
557
self.do_insert_stream_request(repository, resume_tokens)
559
def do_insert_stream_request(self, repository, resume_tokens):
560
tokens = [token for token in resume_tokens.split(' ') if token]
562
self.repository = repository
563
self.queue = Queue.Queue()
564
self.insert_thread = threading.Thread(target=self._inserter_thread)
565
self.insert_thread.start()
567
def do_chunk(self, body_stream_chunk):
568
self.queue.put(body_stream_chunk)
570
def _inserter_thread(self):
572
src_format, stream = _byte_stream_to_stream(
573
self.blocking_byte_stream())
574
self.insert_result = self.repository._get_sink().insert_stream(
575
stream, src_format, self.tokens)
576
self.insert_ok = True
578
self.insert_exception = sys.exc_info()
579
self.insert_ok = False
581
def blocking_byte_stream(self):
583
bytes = self.queue.get()
584
if bytes is StopIteration:
590
self.queue.put(StopIteration)
591
if self.insert_thread is not None:
592
self.insert_thread.join()
593
if not self.insert_ok:
594
exc_info = self.insert_exception
595
raise exc_info[0], exc_info[1], exc_info[2]
596
write_group_tokens, missing_keys = self.insert_result
597
if write_group_tokens or missing_keys:
598
# bzip needed? missing keys should typically be a small set.
599
# Should this be a streaming body response ?
600
missing_keys = sorted(missing_keys)
601
bytes = bencode.bencode((write_group_tokens, missing_keys))
602
self.repository.unlock()
603
return SuccessfulSmartServerResponse(('missing-basis', bytes))
605
self.repository.unlock()
606
return SuccessfulSmartServerResponse(('ok', ))
609
class SmartServerRepositoryInsertStream(SmartServerRepositoryInsertStreamLocked):
610
"""Insert a record stream from a RemoteSink into an unlocked repository.
612
This is the same as SmartServerRepositoryInsertStreamLocked, except it
613
takes no lock_tokens; i.e. it works with an unlocked (or lock-free, e.g.
614
like pack format) repository.
619
def do_repository_request(self, repository, resume_tokens):
620
"""StreamSink.insert_stream for a remote repository."""
621
repository.lock_write()
622
self.do_insert_stream_request(repository, resume_tokens)
added
removed
bzrlib/smart/repository.py
