1
# Copyright (C) 2005, 2006, 2007, 2008, 2009 Canonical Ltd
1
# Copyright (C) 2005-2010 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
864
class Repository(_RelockDebugMixin):
867
class Repository(_RelockDebugMixin, bzrdir.ControlComponent):
865
868
"""Repository holding history for one or more branches.
867
870
The repository holds and retrieves historical information including
1026
1029
:seealso: add_inventory, for the contract.
1028
inv_lines = self._serialise_inventory_to_lines(inv)
1031
inv_lines = self._serializer.write_inventory_to_lines(inv)
1029
1032
return self._inventory_add_lines(revision_id, parents,
1030
1033
inv_lines, check_content=False)
1238
1241
"""Check a single text from this repository."""
1239
1242
if kind == 'inventories':
1240
1243
rev_id = record.key[0]
1241
inv = self.deserialise_inventory(rev_id,
1244
inv = self._deserialise_inventory(rev_id,
1242
1245
record.get_bytes_as('fulltext'))
1243
1246
if last_object is not None:
1244
1247
delta = inv._make_delta(last_object)
1289
1292
:param _format: The format of the repository on disk.
1290
1293
: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.
1296
1298
super(Repository, self).__init__()
1297
1299
self._format = _format
1298
1300
# the following are part of the public API for Repository:
1304
1306
self._reconcile_does_inventory_gc = True
1305
1307
self._reconcile_fixes_text_parents = False
1306
1308
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()
1312
1309
self._write_group = None
1313
1310
# Additional places to query for data.
1314
1311
self._fallback_repositories = []
1315
1312
# An InventoryEntry cache, used during deserialization
1316
1313
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
1318
1326
def __repr__(self):
1319
1327
if self._fallback_repositories:
1386
1394
locked = self.is_locked()
1387
1395
result = self.control_files.lock_write(token=token)
1397
self._warn_if_deprecated()
1389
1398
self._note_lock('w')
1390
1399
for repo in self._fallback_repositories:
1391
1400
# Writes don't affect fallback repos
1467
1477
# now gather global repository information
1468
1478
# XXX: This is available for many repos regardless of listability.
1469
if self.bzrdir.root_transport.listable():
1479
if self.user_transport.listable():
1470
1480
# XXX: do we want to __define len__() ?
1471
1481
# Maybe the versionedfiles object should provide a different
1472
1482
# method to get the number of keys.
1482
1492
:param using: If True, list only branches using this repository.
1484
1494
if using and not self.is_shared():
1486
return [self.bzrdir.open_branch()]
1487
except errors.NotBranchError:
1495
return self.bzrdir.list_branches()
1489
1496
class Evaluator(object):
1491
1498
def __init__(self):
1500
1507
except errors.NoRepositoryPresent:
1503
return False, (None, repository)
1510
return False, ([], repository)
1504
1511
self.first_call = False
1506
value = (bzrdir.open_branch(), None)
1507
except errors.NotBranchError:
1508
value = (None, None)
1512
value = (bzrdir.list_branches(), None)
1509
1513
return True, value
1512
for branch, repository in bzrdir.BzrDir.find_bzrdirs(
1513
self.bzrdir.root_transport, evaluate=Evaluator()):
1514
if branch is not None:
1515
branches.append(branch)
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)
1516
1520
if not using and repository is not None:
1517
branches.extend(repository.find_branches())
1521
ret.extend(repository.find_branches())
1520
1524
@needs_read_lock
1521
1525
def search_missing_revision_ids(self, other, revision_id=None, find_ghosts=True):
1899
1903
rev = self._serializer.read_revision_from_string(text)
1900
1904
yield (revid, rev)
1903
def get_revision_xml(self, revision_id):
1904
# TODO: jam 20070210 This shouldn't be necessary since get_revision
1905
# would have already do it.
1906
# TODO: jam 20070210 Just use _serializer.write_revision_to_string()
1907
# TODO: this can't just be replaced by:
1908
# return self._serializer.write_revision_to_string(
1909
# self.get_revision(revision_id))
1910
# as cStringIO preservers the encoding unlike write_revision_to_string
1911
# or some other call down the path.
1912
rev = self.get_revision(revision_id)
1913
rev_tmp = cStringIO.StringIO()
1914
# the current serializer..
1915
self._serializer.write_revision(rev, rev_tmp)
1917
return rev_tmp.getvalue()
1919
1906
def get_deltas_for_revisions(self, revisions, specific_fileids=None):
1920
1907
"""Produce a generator of revision deltas.
2164
2151
selected_keys = set((revid,) for revid in revision_ids)
2165
2152
w = _inv_weave or self.inventories
2166
pb = ui.ui_factory.nested_progress_bar()
2168
return self._find_file_ids_from_xml_inventory_lines(
2169
w.iter_lines_added_or_present_in_keys(
2170
selected_keys, pb=pb),
2153
return self._find_file_ids_from_xml_inventory_lines(
2154
w.iter_lines_added_or_present_in_keys(
2155
selected_keys, pb=None),
2175
2158
def iter_files_bytes(self, desired_files):
2176
2159
"""Iterate through file versions.
2386
2369
"""single-document based inventory iteration."""
2387
2370
inv_xmls = self._iter_inventory_xmls(revision_ids, ordering)
2388
2371
for text, revision_id in inv_xmls:
2389
yield self.deserialise_inventory(revision_id, text)
2372
yield self._deserialise_inventory(revision_id, text)
2391
2374
def _iter_inventory_xmls(self, revision_ids, ordering):
2392
2375
if ordering is None:
2424
2407
next_key = None
2427
def deserialise_inventory(self, revision_id, xml):
2410
def _deserialise_inventory(self, revision_id, xml):
2428
2411
"""Transform the xml into an inventory object.
2430
2413
:param revision_id: The expected revision id of the inventory.
2431
2414
:param xml: A serialised inventory.
2433
2416
result = self._serializer.read_inventory_from_string(xml, revision_id,
2434
entry_cache=self._inventory_entry_cache)
2417
entry_cache=self._inventory_entry_cache,
2418
return_from_cache=self._safe_to_return_from_cache)
2435
2419
if result.revision_id != revision_id:
2436
2420
raise AssertionError('revision id mismatch %s != %s' % (
2437
2421
result.revision_id, revision_id))
2440
def serialise_inventory(self, inv):
2441
return self._serializer.write_inventory_to_string(inv)
2443
def _serialise_inventory_to_lines(self, inv):
2444
return self._serializer.write_inventory_to_lines(inv)
2446
2424
def get_serializer_format(self):
2447
2425
return self._serializer.format_num
2449
2427
@needs_read_lock
2450
def get_inventory_xml(self, revision_id):
2451
"""Get inventory XML as a file object."""
2428
def _get_inventory_xml(self, revision_id):
2429
"""Get serialized inventory as a string."""
2452
2430
texts = self._iter_inventory_xmls([revision_id], 'unordered')
2454
2432
text, revision_id = texts.next()
2456
2434
raise errors.HistoryMissing(self, 'inventory', revision_id)
2460
def get_inventory_sha1(self, revision_id):
2461
"""Return the sha1 hash of the inventory entry
2463
return self.get_revision(revision_id).inventory_sha1
2465
2437
def get_rev_id_for_revno(self, revno, known_pair):
2466
2438
"""Return the revision id of a revno, given a later (revno, revid)
2467
2439
pair in the same history.
2519
2491
next_id = parents[0]
2522
def get_revision_inventory(self, revision_id):
2523
"""Return inventory of a past revision."""
2524
# TODO: Unify this with get_inventory()
2525
# bzr 0.0.6 and later imposes the constraint that the inventory_id
2526
# must be the same as its revision, so this is trivial.
2527
if revision_id is None:
2528
# This does not make sense: if there is no revision,
2529
# then it is the current tree inventory surely ?!
2530
# and thus get_root_id() is something that looks at the last
2531
# commit on the branch, and the get_root_id is an inventory check.
2532
raise NotImplementedError
2533
# return Inventory(self.get_root_id())
2535
return self.get_inventory(revision_id)
2537
2493
def is_shared(self):
2538
2494
"""Return True if this repository is flagged as a shared repository."""
2539
2495
raise NotImplementedError(self.is_shared)
2573
2529
return RevisionTree(self, Inventory(root_id=None),
2574
2530
_mod_revision.NULL_REVISION)
2576
inv = self.get_revision_inventory(revision_id)
2532
inv = self.get_inventory(revision_id)
2577
2533
return RevisionTree(self, inv, revision_id)
2579
2535
def revision_trees(self, revision_ids):
2632
2588
keys = tsort.topo_sort(parent_map)
2633
2589
return [None] + list(keys)
2635
def pack(self, hint=None):
2591
def pack(self, hint=None, clean_obsolete_packs=False):
2636
2592
"""Compress the data within the repository.
2638
2594
This operation only makes sense for some repository types. For other
2648
2604
obtained from the result of commit_write_group(). Out of
2649
2605
date hints are simply ignored, because concurrent operations
2650
2606
can obsolete them rapidly.
2608
:param clean_obsolete_packs: Clean obsolete packs immediately after
2653
2612
def get_transaction(self):
2678
2637
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)
2681
2649
def get_graph(self, other_repository=None):
2682
2650
"""Return the graph walker for this repository format"""
2683
2651
parents_provider = self._make_parents_provider()
2778
2746
result.check(callback_refs)
2781
def _warn_if_deprecated(self):
2749
def _warn_if_deprecated(self, branch=None):
2782
2750
global _deprecation_warning_done
2783
2751
if _deprecation_warning_done:
2785
_deprecation_warning_done = True
2786
warning("Format %s for %s is deprecated - please use 'bzr upgrade' to get better performance"
2787
% (self._format, self.bzrdir.transport.base))
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
2766
def supports_rich_root(self):
2790
2767
return self._format.rich_root_data
3073
3050
pack_compresses = False
3074
3051
# Does the repository inventory storage understand references to trees?
3075
3052
supports_tree_reference = None
3053
# Is the format experimental ?
3054
experimental = False
3078
return "<%s>" % self.__class__.__name__
3057
return "%s()" % self.__class__.__name__
3080
3059
def __eq__(self, other):
3081
3060
# format objects are generally stateless
3200
3179
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)
3203
3191
class MetaDirRepositoryFormat(RepositoryFormat):
3204
3192
"""Common base class for the new repositories using the metadir layout."""
3410
3398
:param revision_id: if None all content is copied, if NULL_REVISION no
3411
3399
content is copied.
3412
:param pb: optional progress bar to use for progress reports. If not
3413
provided a default one will be created.
3416
f = _mod_fetch.RepoFetcher(to_repository=self.target,
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,
3417
3411
from_repository=self.source,
3418
3412
last_revision=revision_id,
3419
3413
fetch_spec=fetch_spec,
3420
pb=pb, find_ghosts=find_ghosts)
3414
find_ghosts=find_ghosts)
3422
3416
def _walk_to_common_revisions(self, revision_ids):
3423
3417
"""Walk out from revision_ids in source to revisions target has.
3840
3834
pending_revisions = []
3841
3835
parent_map = self.source.get_parent_map(revision_ids)
3842
3836
self._fetch_parent_invs_for_stacking(parent_map, cache)
3837
self.source._safe_to_return_from_cache = True
3843
3838
for tree in self.source.revision_trees(revision_ids):
3844
3839
# Find a inventory delta for this revision.
3845
3840
# Find text entries that need to be copied, too.
3893
3888
pending_revisions.append(revision)
3894
3889
cache[current_revision_id] = tree
3895
3890
basis_id = current_revision_id
3891
self.source._safe_to_return_from_cache = False
3896
3892
# Copy file texts
3897
3893
from_texts = self.source.texts
3898
3894
to_texts = self.target.texts
3977
3973
basis_id = self._fetch_batch(batch, basis_id, cache,
3978
3974
a_graph=a_graph)
3976
self.source._safe_to_return_from_cache = False
3980
3977
self.target.abort_write_group()
3994
3991
"""See InterRepository.fetch()."""
3995
3992
if fetch_spec is not None:
3996
3993
raise AssertionError("Not implemented yet...")
3994
ui.ui_factory.warn_experimental_format_fetch(self)
3997
3995
if (not self.source.supports_rich_root()
3998
3996
and self.target.supports_rich_root()):
3999
3997
self._converting_to_rich_root = True
4000
3998
self._revision_id_to_root_id = {}
4002
4000
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)
4003
4006
revision_ids = self.target.search_missing_revision_ids(self.source,
4004
4007
revision_id, find_ghosts=find_ghosts).get_keys()
4005
4008
if not revision_ids:
4074
4077
:param to_convert: The disk object to convert.
4075
4078
:param pb: a progress bar to use for progress information.
4080
pb = ui.ui_factory.nested_progress_bar()
4080
4083
# this is only useful with metadir layouts - separated repo content.
4081
4084
# trigger an assertion if not such
4082
4085
repo._format.get_format_string()
4083
4086
self.repo_dir = repo.bzrdir
4084
self.step('Moving repository to repository.backup')
4087
pb.update('Moving repository to repository.backup')
4085
4088
self.repo_dir.transport.move('repository', 'repository.backup')
4086
4089
backup_transport = self.repo_dir.transport.clone('repository.backup')
4087
4090
repo._format.check_conversion_target(self.target_format)
4088
4091
self.source_repo = repo._format.open(self.repo_dir,
4090
4093
_override_transport=backup_transport)
4091
self.step('Creating new repository')
4094
pb.update('Creating new repository')
4092
4095
converted = self.target_format.initialize(self.repo_dir,
4093
4096
self.source_repo.is_shared())
4094
4097
converted.lock_write()
4096
self.step('Copying content')
4099
pb.update('Copying content')
4097
4100
self.source_repo.copy_content_into(converted)
4099
4102
converted.unlock()
4100
self.step('Deleting old repository content')
4103
pb.update('Deleting old repository content')
4101
4104
self.repo_dir.transport.delete_tree('repository.backup')
4102
4105
ui.ui_factory.note('repository converted')
4104
def step(self, message):
4105
"""Update the pb by a step."""
4107
self.pb.update(message, self.count, self.total)
4110
4109
_unescape_map = {
4603
4602
def _get_convertable_inventory_stream(self, revision_ids,
4604
4603
delta_versus_null=False):
4605
# The source is using CHKs, but the target either doesn't or it has a
4606
# different serializer. The StreamSink code expects to be able to
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
4607
4608
# convert on the target, so we need to put bytes-on-the-wire that can
4608
4609
# be converted. That means inventory deltas (if the remote is <1.19,
4609
4610
# RemoteStreamSink will fallback to VFS to insert the deltas).
added
removed
bzrlib/repository.py
