1
# Copyright (C) 2005-2010 Canonical Ltd
1
# Copyright (C) 2005, 2006, 2007, 2008, 2009 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
867
class Repository(_RelockDebugMixin, bzrdir.ControlComponent):
865
class Repository(_RelockDebugMixin):
868
866
"""Repository holding history for one or more branches.
870
868
The repository holds and retrieves historical information including
1029
1027
:seealso: add_inventory, for the contract.
1031
inv_lines = self._serializer.write_inventory_to_lines(inv)
1029
inv_lines = self._serialise_inventory_to_lines(inv)
1032
1030
return self._inventory_add_lines(revision_id, parents,
1033
1031
inv_lines, check_content=False)
1241
1239
"""Check a single text from this repository."""
1242
1240
if kind == 'inventories':
1243
1241
rev_id = record.key[0]
1244
inv = self._deserialise_inventory(rev_id,
1242
inv = self.deserialise_inventory(rev_id,
1245
1243
record.get_bytes_as('fulltext'))
1246
1244
if last_object is not None:
1247
1245
delta = inv._make_delta(last_object)
1292
1290
:param _format: The format of the repository on disk.
1293
1291
:param a_bzrdir: The BzrDir of the repository.
1293
In the future we will have a single api for all stores for
1294
getting file texts, inventories and revisions, then
1295
this construct will accept instances of those things.
1295
# In the future we will have a single api for all stores for
1296
# getting file texts, inventories and revisions, then
1297
# this construct will accept instances of those things.
1298
1297
super(Repository, self).__init__()
1299
1298
self._format = _format
1300
1299
# the following are part of the public API for Repository:
1311
1310
self._fallback_repositories = []
1312
1311
# An InventoryEntry cache, used during deserialization
1313
1312
self._inventory_entry_cache = fifo_cache.FIFOCache(10*1024)
1314
# Is it safe to return inventory entries directly from the entry cache,
1315
# rather copying them?
1316
self._safe_to_return_from_cache = False
1319
def user_transport(self):
1320
return self.bzrdir.user_transport
1323
def control_transport(self):
1324
return self._transport
1326
1314
def __repr__(self):
1327
1315
if self._fallback_repositories:
1477
1465
# now gather global repository information
1478
1466
# XXX: This is available for many repos regardless of listability.
1479
if self.user_transport.listable():
1467
if self.bzrdir.root_transport.listable():
1480
1468
# XXX: do we want to __define len__() ?
1481
1469
# Maybe the versionedfiles object should provide a different
1482
1470
# method to get the number of keys.
1492
1480
:param using: If True, list only branches using this repository.
1494
1482
if using and not self.is_shared():
1495
return self.bzrdir.list_branches()
1484
return [self.bzrdir.open_branch()]
1485
except errors.NotBranchError:
1496
1487
class Evaluator(object):
1498
1489
def __init__(self):
1507
1498
except errors.NoRepositoryPresent:
1510
return False, ([], repository)
1501
return False, (None, repository)
1511
1502
self.first_call = False
1512
value = (bzrdir.list_branches(), None)
1504
value = (bzrdir.open_branch(), None)
1505
except errors.NotBranchError:
1506
value = (None, None)
1513
1507
return True, value
1516
for branches, repository in bzrdir.BzrDir.find_bzrdirs(
1517
self.user_transport, evaluate=Evaluator()):
1518
if branches is not None:
1519
ret.extend(branches)
1510
for branch, repository in bzrdir.BzrDir.find_bzrdirs(
1511
self.bzrdir.root_transport, evaluate=Evaluator()):
1512
if branch is not None:
1513
branches.append(branch)
1520
1514
if not using and repository is not None:
1521
ret.extend(repository.find_branches())
1515
branches.extend(repository.find_branches())
1524
1518
@needs_read_lock
1525
1519
def search_missing_revision_ids(self, other, revision_id=None, find_ghosts=True):
1903
1897
rev = self._serializer.read_revision_from_string(text)
1904
1898
yield (revid, rev)
1901
def get_revision_xml(self, revision_id):
1902
# TODO: jam 20070210 This shouldn't be necessary since get_revision
1903
# would have already do it.
1904
# TODO: jam 20070210 Just use _serializer.write_revision_to_string()
1905
# TODO: this can't just be replaced by:
1906
# return self._serializer.write_revision_to_string(
1907
# self.get_revision(revision_id))
1908
# as cStringIO preservers the encoding unlike write_revision_to_string
1909
# or some other call down the path.
1910
rev = self.get_revision(revision_id)
1911
rev_tmp = cStringIO.StringIO()
1912
# the current serializer..
1913
self._serializer.write_revision(rev, rev_tmp)
1915
return rev_tmp.getvalue()
1906
1917
def get_deltas_for_revisions(self, revisions, specific_fileids=None):
1907
1918
"""Produce a generator of revision deltas.
2151
2162
selected_keys = set((revid,) for revid in revision_ids)
2152
2163
w = _inv_weave or self.inventories
2153
return self._find_file_ids_from_xml_inventory_lines(
2154
w.iter_lines_added_or_present_in_keys(
2155
selected_keys, pb=None),
2164
pb = ui.ui_factory.nested_progress_bar()
2166
return self._find_file_ids_from_xml_inventory_lines(
2167
w.iter_lines_added_or_present_in_keys(
2168
selected_keys, pb=pb),
2158
2173
def iter_files_bytes(self, desired_files):
2159
2174
"""Iterate through file versions.
2369
2384
"""single-document based inventory iteration."""
2370
2385
inv_xmls = self._iter_inventory_xmls(revision_ids, ordering)
2371
2386
for text, revision_id in inv_xmls:
2372
yield self._deserialise_inventory(revision_id, text)
2387
yield self.deserialise_inventory(revision_id, text)
2374
2389
def _iter_inventory_xmls(self, revision_ids, ordering):
2375
2390
if ordering is None:
2407
2422
next_key = None
2410
def _deserialise_inventory(self, revision_id, xml):
2425
def deserialise_inventory(self, revision_id, xml):
2411
2426
"""Transform the xml into an inventory object.
2413
2428
:param revision_id: The expected revision id of the inventory.
2414
2429
:param xml: A serialised inventory.
2416
2431
result = self._serializer.read_inventory_from_string(xml, revision_id,
2417
entry_cache=self._inventory_entry_cache,
2418
return_from_cache=self._safe_to_return_from_cache)
2432
entry_cache=self._inventory_entry_cache)
2419
2433
if result.revision_id != revision_id:
2420
2434
raise AssertionError('revision id mismatch %s != %s' % (
2421
2435
result.revision_id, revision_id))
2438
def serialise_inventory(self, inv):
2439
return self._serializer.write_inventory_to_string(inv)
2441
def _serialise_inventory_to_lines(self, inv):
2442
return self._serializer.write_inventory_to_lines(inv)
2424
2444
def get_serializer_format(self):
2425
2445
return self._serializer.format_num
2427
2447
@needs_read_lock
2428
def _get_inventory_xml(self, revision_id):
2429
"""Get serialized inventory as a string."""
2448
def get_inventory_xml(self, revision_id):
2449
"""Get inventory XML as a file object."""
2430
2450
texts = self._iter_inventory_xmls([revision_id], 'unordered')
2432
2452
text, revision_id = texts.next()
2434
2454
raise errors.HistoryMissing(self, 'inventory', revision_id)
2458
def get_inventory_sha1(self, revision_id):
2459
"""Return the sha1 hash of the inventory entry
2461
return self.get_revision(revision_id).inventory_sha1
2437
2463
def get_rev_id_for_revno(self, revno, known_pair):
2438
2464
"""Return the revision id of a revno, given a later (revno, revid)
2439
2465
pair in the same history.
2491
2517
next_id = parents[0]
2520
def get_revision_inventory(self, revision_id):
2521
"""Return inventory of a past revision."""
2522
# TODO: Unify this with get_inventory()
2523
# bzr 0.0.6 and later imposes the constraint that the inventory_id
2524
# must be the same as its revision, so this is trivial.
2525
if revision_id is None:
2526
# This does not make sense: if there is no revision,
2527
# then it is the current tree inventory surely ?!
2528
# and thus get_root_id() is something that looks at the last
2529
# commit on the branch, and the get_root_id is an inventory check.
2530
raise NotImplementedError
2531
# return Inventory(self.get_root_id())
2533
return self.get_inventory(revision_id)
2493
2535
def is_shared(self):
2494
2536
"""Return True if this repository is flagged as a shared repository."""
2495
2537
raise NotImplementedError(self.is_shared)
2529
2571
return RevisionTree(self, Inventory(root_id=None),
2530
2572
_mod_revision.NULL_REVISION)
2532
inv = self.get_inventory(revision_id)
2574
inv = self.get_revision_inventory(revision_id)
2533
2575
return RevisionTree(self, inv, revision_id)
2535
2577
def revision_trees(self, revision_ids):
2588
2630
keys = tsort.topo_sort(parent_map)
2589
2631
return [None] + list(keys)
2591
def pack(self, hint=None, clean_obsolete_packs=False):
2633
def pack(self, hint=None):
2592
2634
"""Compress the data within the repository.
2594
2636
This operation only makes sense for some repository types. For other
2604
2646
obtained from the result of commit_write_group(). Out of
2605
2647
date hints are simply ignored, because concurrent operations
2606
2648
can obsolete them rapidly.
2608
:param clean_obsolete_packs: Clean obsolete packs immediately after
2612
2651
def get_transaction(self):
2637
2676
def _make_parents_provider(self):
2641
def get_known_graph_ancestry(self, revision_ids):
2642
"""Return the known graph for a set of revision ids and their ancestors.
2644
st = static_tuple.StaticTuple
2645
revision_keys = [st(r_id).intern() for r_id in revision_ids]
2646
known_graph = self.revisions.get_known_graph_ancestry(revision_keys)
2647
return graph.GraphThunkIdsToKeys(known_graph)
2649
2679
def get_graph(self, other_repository=None):
2650
2680
"""Return the graph walker for this repository format"""
2651
2681
parents_provider = self._make_parents_provider()
3050
3080
pack_compresses = False
3051
3081
# Does the repository inventory storage understand references to trees?
3052
3082
supports_tree_reference = None
3053
# Is the format experimental ?
3054
experimental = False
3057
return "%s()" % self.__class__.__name__
3085
return "<%s>" % self.__class__.__name__
3059
3087
def __eq__(self, other):
3060
3088
# format objects are generally stateless
3075
3103
transport = a_bzrdir.get_repository_transport(None)
3076
format_string = transport.get_bytes("format")
3104
format_string = transport.get("format").read()
3077
3105
return format_registry.get(format_string)
3078
3106
except errors.NoSuchFile:
3079
3107
raise errors.NoRepositoryPresent(a_bzrdir)
3179
3207
raise NotImplementedError(self.open)
3181
def _run_post_repo_init_hooks(self, repository, a_bzrdir, shared):
3182
from bzrlib.bzrdir import BzrDir, RepoInitHookParams
3183
hooks = BzrDir.hooks['post_repo_init']
3186
params = RepoInitHookParams(repository, self, a_bzrdir, shared)
3191
3210
class MetaDirRepositoryFormat(RepositoryFormat):
3192
3211
"""Common base class for the new repositories using the metadir layout."""
3398
3417
:param revision_id: if None all content is copied, if NULL_REVISION no
3399
3418
content is copied.
3419
:param pb: optional progress bar to use for progress reports. If not
3420
provided a default one will be created.
3403
ui.ui_factory.warn_experimental_format_fetch(self)
3404
from bzrlib.fetch import RepoFetcher
3405
# See <https://launchpad.net/bugs/456077> asking for a warning here
3406
if self.source._format.network_name() != self.target._format.network_name():
3407
ui.ui_factory.show_user_warning('cross_format_fetch',
3408
from_format=self.source._format,
3409
to_format=self.target._format)
3410
f = RepoFetcher(to_repository=self.target,
3423
f = _mod_fetch.RepoFetcher(to_repository=self.target,
3411
3424
from_repository=self.source,
3412
3425
last_revision=revision_id,
3413
3426
fetch_spec=fetch_spec,
3414
find_ghosts=find_ghosts)
3427
pb=pb, find_ghosts=find_ghosts)
3416
3429
def _walk_to_common_revisions(self, revision_ids):
3417
3430
"""Walk out from revision_ids in source to revisions target has.
3834
3847
pending_revisions = []
3835
3848
parent_map = self.source.get_parent_map(revision_ids)
3836
3849
self._fetch_parent_invs_for_stacking(parent_map, cache)
3837
self.source._safe_to_return_from_cache = True
3838
3850
for tree in self.source.revision_trees(revision_ids):
3839
3851
# Find a inventory delta for this revision.
3840
3852
# Find text entries that need to be copied, too.
3888
3900
pending_revisions.append(revision)
3889
3901
cache[current_revision_id] = tree
3890
3902
basis_id = current_revision_id
3891
self.source._safe_to_return_from_cache = False
3892
3903
# Copy file texts
3893
3904
from_texts = self.source.texts
3894
3905
to_texts = self.target.texts
3973
3984
basis_id = self._fetch_batch(batch, basis_id, cache,
3974
3985
a_graph=a_graph)
3976
self.source._safe_to_return_from_cache = False
3977
3987
self.target.abort_write_group()
3991
4001
"""See InterRepository.fetch()."""
3992
4002
if fetch_spec is not None:
3993
4003
raise AssertionError("Not implemented yet...")
3994
ui.ui_factory.warn_experimental_format_fetch(self)
3995
4004
if (not self.source.supports_rich_root()
3996
4005
and self.target.supports_rich_root()):
3997
4006
self._converting_to_rich_root = True
3998
4007
self._revision_id_to_root_id = {}
4000
4009
self._converting_to_rich_root = False
4001
# See <https://launchpad.net/bugs/456077> asking for a warning here
4002
if self.source._format.network_name() != self.target._format.network_name():
4003
ui.ui_factory.show_user_warning('cross_format_fetch',
4004
from_format=self.source._format,
4005
to_format=self.target._format)
4006
4010
revision_ids = self.target.search_missing_revision_ids(self.source,
4007
4011
revision_id, find_ghosts=find_ghosts).get_keys()
4008
4012
if not revision_ids:
4077
4081
:param to_convert: The disk object to convert.
4078
4082
:param pb: a progress bar to use for progress information.
4080
pb = ui.ui_factory.nested_progress_bar()
4083
4087
# this is only useful with metadir layouts - separated repo content.
4084
4088
# trigger an assertion if not such
4085
4089
repo._format.get_format_string()
4086
4090
self.repo_dir = repo.bzrdir
4087
pb.update('Moving repository to repository.backup')
4091
self.step('Moving repository to repository.backup')
4088
4092
self.repo_dir.transport.move('repository', 'repository.backup')
4089
4093
backup_transport = self.repo_dir.transport.clone('repository.backup')
4090
4094
repo._format.check_conversion_target(self.target_format)
4091
4095
self.source_repo = repo._format.open(self.repo_dir,
4093
4097
_override_transport=backup_transport)
4094
pb.update('Creating new repository')
4098
self.step('Creating new repository')
4095
4099
converted = self.target_format.initialize(self.repo_dir,
4096
4100
self.source_repo.is_shared())
4097
4101
converted.lock_write()
4099
pb.update('Copying content')
4103
self.step('Copying content')
4100
4104
self.source_repo.copy_content_into(converted)
4102
4106
converted.unlock()
4103
pb.update('Deleting old repository content')
4107
self.step('Deleting old repository content')
4104
4108
self.repo_dir.transport.delete_tree('repository.backup')
4105
4109
ui.ui_factory.note('repository converted')
4111
def step(self, message):
4112
"""Update the pb by a step."""
4114
self.pb.update(message, self.count, self.total)
4109
4117
_unescape_map = {
4602
4610
def _get_convertable_inventory_stream(self, revision_ids,
4603
4611
delta_versus_null=False):
4604
# The two formats are sufficiently different that there is no fast
4605
# path, so we need to send just inventorydeltas, which any
4606
# sufficiently modern client can insert into any repository.
4607
# The StreamSink code expects to be able to
4612
# The source is using CHKs, but the target either doesn't or it has a
4613
# different serializer. The StreamSink code expects to be able to
4608
4614
# convert on the target, so we need to put bytes-on-the-wire that can
4609
4615
# be converted. That means inventory deltas (if the remote is <1.19,
4610
4616
# RemoteStreamSink will fallback to VFS to insert the deltas).
added
removed
bzrlib/repository.py
