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):
864
class Repository(_RelockDebugMixin):
868
865
"""Repository holding history for one or more branches.
870
867
The repository holds and retrieves historical information including
1029
1026
:seealso: add_inventory, for the contract.
1031
inv_lines = self._serializer.write_inventory_to_lines(inv)
1028
inv_lines = self._serialise_inventory_to_lines(inv)
1032
1029
return self._inventory_add_lines(revision_id, parents,
1033
1030
inv_lines, check_content=False)
1241
1238
"""Check a single text from this repository."""
1242
1239
if kind == 'inventories':
1243
1240
rev_id = record.key[0]
1244
inv = self._deserialise_inventory(rev_id,
1241
inv = self.deserialise_inventory(rev_id,
1245
1242
record.get_bytes_as('fulltext'))
1246
1243
if last_object is not None:
1247
1244
delta = inv._make_delta(last_object)
1292
1289
:param _format: The format of the repository on disk.
1293
1290
:param a_bzrdir: The BzrDir of the repository.
1292
In the future we will have a single api for all stores for
1293
getting file texts, inventories and revisions, then
1294
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
1296
super(Repository, self).__init__()
1299
1297
self._format = _format
1300
1298
# the following are part of the public API for Repository:
1306
1304
self._reconcile_does_inventory_gc = True
1307
1305
self._reconcile_fixes_text_parents = False
1308
1306
self._reconcile_backsup_inventory = True
1307
# not right yet - should be more semantically clear ?
1309
# TODO: make sure to construct the right store classes, etc, depending
1310
# on whether escaping is required.
1311
self._warn_if_deprecated()
1309
1312
self._write_group = None
1310
1313
# Additional places to query for data.
1311
1314
self._fallback_repositories = []
1315
1318
# rather copying them?
1316
1319
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
1321
def __repr__(self):
1327
1322
if self._fallback_repositories:
1328
1323
return '%s(%r, fallback_repositories=%r)' % (
1394
1389
locked = self.is_locked()
1395
1390
result = self.control_files.lock_write(token=token)
1397
self._warn_if_deprecated()
1398
1392
self._note_lock('w')
1399
1393
for repo in self._fallback_repositories:
1400
1394
# Writes don't affect fallback repos
1477
1470
# now gather global repository information
1478
1471
# XXX: This is available for many repos regardless of listability.
1479
if self.user_transport.listable():
1472
if self.bzrdir.root_transport.listable():
1480
1473
# XXX: do we want to __define len__() ?
1481
1474
# Maybe the versionedfiles object should provide a different
1482
1475
# method to get the number of keys.
1492
1485
:param using: If True, list only branches using this repository.
1494
1487
if using and not self.is_shared():
1495
return self.bzrdir.list_branches()
1489
return [self.bzrdir.open_branch()]
1490
except errors.NotBranchError:
1496
1492
class Evaluator(object):
1498
1494
def __init__(self):
1507
1503
except errors.NoRepositoryPresent:
1510
return False, ([], repository)
1506
return False, (None, repository)
1511
1507
self.first_call = False
1512
value = (bzrdir.list_branches(), None)
1509
value = (bzrdir.open_branch(), None)
1510
except errors.NotBranchError:
1511
value = (None, None)
1513
1512
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)
1515
for branch, repository in bzrdir.BzrDir.find_bzrdirs(
1516
self.bzrdir.root_transport, evaluate=Evaluator()):
1517
if branch is not None:
1518
branches.append(branch)
1520
1519
if not using and repository is not None:
1521
ret.extend(repository.find_branches())
1520
branches.extend(repository.find_branches())
1524
1523
@needs_read_lock
1525
1524
def search_missing_revision_ids(self, other, revision_id=None, find_ghosts=True):
1903
1902
rev = self._serializer.read_revision_from_string(text)
1904
1903
yield (revid, rev)
1906
def get_revision_xml(self, revision_id):
1907
# TODO: jam 20070210 This shouldn't be necessary since get_revision
1908
# would have already do it.
1909
# TODO: jam 20070210 Just use _serializer.write_revision_to_string()
1910
# TODO: this can't just be replaced by:
1911
# return self._serializer.write_revision_to_string(
1912
# self.get_revision(revision_id))
1913
# as cStringIO preservers the encoding unlike write_revision_to_string
1914
# or some other call down the path.
1915
rev = self.get_revision(revision_id)
1916
rev_tmp = cStringIO.StringIO()
1917
# the current serializer..
1918
self._serializer.write_revision(rev, rev_tmp)
1920
return rev_tmp.getvalue()
1906
1922
def get_deltas_for_revisions(self, revisions, specific_fileids=None):
1907
1923
"""Produce a generator of revision deltas.
2151
2167
selected_keys = set((revid,) for revid in revision_ids)
2152
2168
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),
2169
pb = ui.ui_factory.nested_progress_bar()
2171
return self._find_file_ids_from_xml_inventory_lines(
2172
w.iter_lines_added_or_present_in_keys(
2173
selected_keys, pb=pb),
2158
2178
def iter_files_bytes(self, desired_files):
2159
2179
"""Iterate through file versions.
2369
2389
"""single-document based inventory iteration."""
2370
2390
inv_xmls = self._iter_inventory_xmls(revision_ids, ordering)
2371
2391
for text, revision_id in inv_xmls:
2372
yield self._deserialise_inventory(revision_id, text)
2392
yield self.deserialise_inventory(revision_id, text)
2374
2394
def _iter_inventory_xmls(self, revision_ids, ordering):
2375
2395
if ordering is None:
2407
2427
next_key = None
2410
def _deserialise_inventory(self, revision_id, xml):
2430
def deserialise_inventory(self, revision_id, xml):
2411
2431
"""Transform the xml into an inventory object.
2413
2433
:param revision_id: The expected revision id of the inventory.
2421
2441
result.revision_id, revision_id))
2444
def serialise_inventory(self, inv):
2445
return self._serializer.write_inventory_to_string(inv)
2447
def _serialise_inventory_to_lines(self, inv):
2448
return self._serializer.write_inventory_to_lines(inv)
2424
2450
def get_serializer_format(self):
2425
2451
return self._serializer.format_num
2427
2453
@needs_read_lock
2428
def _get_inventory_xml(self, revision_id):
2429
"""Get serialized inventory as a string."""
2454
def get_inventory_xml(self, revision_id):
2455
"""Get inventory XML as a file object."""
2430
2456
texts = self._iter_inventory_xmls([revision_id], 'unordered')
2432
2458
text, revision_id = texts.next()
2434
2460
raise errors.HistoryMissing(self, 'inventory', revision_id)
2464
def get_inventory_sha1(self, revision_id):
2465
"""Return the sha1 hash of the inventory entry
2467
return self.get_revision(revision_id).inventory_sha1
2437
2469
def get_rev_id_for_revno(self, revno, known_pair):
2438
2470
"""Return the revision id of a revno, given a later (revno, revid)
2439
2471
pair in the same history.
2491
2523
next_id = parents[0]
2526
def get_revision_inventory(self, revision_id):
2527
"""Return inventory of a past revision."""
2528
# TODO: Unify this with get_inventory()
2529
# bzr 0.0.6 and later imposes the constraint that the inventory_id
2530
# must be the same as its revision, so this is trivial.
2531
if revision_id is None:
2532
# This does not make sense: if there is no revision,
2533
# then it is the current tree inventory surely ?!
2534
# and thus get_root_id() is something that looks at the last
2535
# commit on the branch, and the get_root_id is an inventory check.
2536
raise NotImplementedError
2537
# return Inventory(self.get_root_id())
2539
return self.get_inventory(revision_id)
2493
2541
def is_shared(self):
2494
2542
"""Return True if this repository is flagged as a shared repository."""
2495
2543
raise NotImplementedError(self.is_shared)
2529
2577
return RevisionTree(self, Inventory(root_id=None),
2530
2578
_mod_revision.NULL_REVISION)
2532
inv = self.get_inventory(revision_id)
2580
inv = self.get_revision_inventory(revision_id)
2533
2581
return RevisionTree(self, inv, revision_id)
2535
2583
def revision_trees(self, revision_ids):
2588
2636
keys = tsort.topo_sort(parent_map)
2589
2637
return [None] + list(keys)
2591
def pack(self, hint=None, clean_obsolete_packs=False):
2639
def pack(self, hint=None):
2592
2640
"""Compress the data within the repository.
2594
2642
This operation only makes sense for some repository types. For other
2604
2652
obtained from the result of commit_write_group(). Out of
2605
2653
date hints are simply ignored, because concurrent operations
2606
2654
can obsolete them rapidly.
2608
:param clean_obsolete_packs: Clean obsolete packs immediately after
2612
2657
def get_transaction(self):
2637
2682
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
2685
def get_graph(self, other_repository=None):
2650
2686
"""Return the graph walker for this repository format"""
2651
2687
parents_provider = self._make_parents_provider()
2746
2782
result.check(callback_refs)
2749
def _warn_if_deprecated(self, branch=None):
2785
def _warn_if_deprecated(self):
2750
2786
global _deprecation_warning_done
2751
2787
if _deprecation_warning_done:
2755
conf = config.GlobalConfig()
2757
conf = branch.get_config()
2758
if conf.suppress_warning('format_deprecation'):
2760
warning("Format %s for %s is deprecated -"
2761
" please use 'bzr upgrade' to get better performance"
2762
% (self._format, self.bzrdir.transport.base))
2764
_deprecation_warning_done = True
2789
_deprecation_warning_done = True
2790
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
2791
% (self._format, self.bzrdir.transport.base))
2766
2793
def supports_rich_root(self):
2767
2794
return self._format.rich_root_data
3050
3077
pack_compresses = False
3051
3078
# Does the repository inventory storage understand references to trees?
3052
3079
supports_tree_reference = None
3053
# Is the format experimental ?
3054
experimental = False
3057
return "%s()" % self.__class__.__name__
3082
return "<%s>" % self.__class__.__name__
3059
3084
def __eq__(self, other):
3060
3085
# format objects are generally stateless
3179
3204
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
3207
class MetaDirRepositoryFormat(RepositoryFormat):
3192
3208
"""Common base class for the new repositories using the metadir layout."""
3398
3414
:param revision_id: if None all content is copied, if NULL_REVISION no
3399
3415
content is copied.
3416
:param pb: optional progress bar to use for progress reports. If not
3417
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,
3420
f = _mod_fetch.RepoFetcher(to_repository=self.target,
3411
3421
from_repository=self.source,
3412
3422
last_revision=revision_id,
3413
3423
fetch_spec=fetch_spec,
3414
find_ghosts=find_ghosts)
3424
pb=pb, find_ghosts=find_ghosts)
3416
3426
def _walk_to_common_revisions(self, revision_ids):
3417
3427
"""Walk out from revision_ids in source to revisions target has.
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
