1 # Copyright 2011-2012 GRNET S.A. All rights reserved.
3 # Redistribution and use in source and binary forms, with or
4 # without modification, are permitted provided that the following
7 # 1. Redistributions of source code must retain the above
8 # copyright notice, this list of conditions and the following
11 # 2. Redistributions in binary form must reproduce the above
12 # copyright notice, this list of conditions and the following
13 # disclaimer in the documentation and/or other materials
14 # provided with the distribution.
16 # THIS SOFTWARE IS PROVIDED BY GRNET S.A. ``AS IS'' AND ANY EXPRESS
17 # OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
18 # WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
19 # PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL GRNET S.A OR
20 # CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
21 # SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
22 # LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF
23 # USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
24 # AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25 # LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
26 # ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
27 # POSSIBILITY OF SUCH DAMAGE.
29 # The views and conclusions contained in the software and
30 # documentation are those of the authors and should not be
31 # interpreted as representing official policies, either expressed
32 # or implied, of GRNET S.A.
37 import uuid as uuidlib
42 from base import DEFAULT_QUOTA, DEFAULT_VERSIONING, NotAllowedError, QuotaError, BaseBackend
44 # Stripped-down version of the HashMap class found in tools.
47 def __init__(self, blocksize, blockhash):
48 super(HashMap, self).__init__()
49 self.blocksize = blocksize
50 self.blockhash = blockhash
52 def _hash_raw(self, v):
53 h = hashlib.new(self.blockhash)
59 return self._hash_raw('')
61 return self.__getitem__(0)
67 h += [('\x00' * len(h[0]))] * (s - len(h))
69 h = [self._hash_raw(h[x] + h[x + 1]) for x in range(0, len(h), 2)]
72 # Default modules and settings.
73 DEFAULT_DB_MODULE = 'pithos.backends.lib.sqlalchemy'
74 DEFAULT_DB_CONNECTION = 'sqlite:///backend.db'
75 DEFAULT_BLOCK_MODULE = 'pithos.backends.lib.hashfiler'
76 DEFAULT_BLOCK_PATH = 'data/'
77 DEFAULT_BLOCK_UMASK = 0o022
78 #DEFAULT_QUEUE_MODULE = 'pithos.backends.lib.rabbitmq'
79 #DEFAULT_QUEUE_CONNECTION = 'rabbitmq://guest:guest@localhost:5672/pithos'
81 QUEUE_MESSAGE_KEY_PREFIX = 'pithos.%s'
82 QUEUE_CLIENT_ID = 'pithos'
83 QUEUE_INSTANCE_ID = '1'
85 ( CLUSTER_NORMAL, CLUSTER_HISTORY, CLUSTER_DELETED ) = range(3)
92 logger = logging.getLogger(__name__)
95 def backend_method(func=None, autocommit=1):
98 return backend_method(func, autocommit)
103 def fn(self, *args, **kw):
104 self.wrapper.execute()
107 ret = func(self, *args, **kw)
108 for m in self.messages:
110 self.wrapper.commit()
113 self.wrapper.rollback()
118 class ModularBackend(BaseBackend):
119 """A modular backend.
121 Uses modules for SQL functions and storage.
124 def __init__(self, db_module=None, db_connection=None,
125 block_module=None, block_path=None, block_umask=None,
126 queue_module=None, queue_connection=None):
127 db_module = db_module or DEFAULT_DB_MODULE
128 db_connection = db_connection or DEFAULT_DB_CONNECTION
129 block_module = block_module or DEFAULT_BLOCK_MODULE
130 block_path = block_path or DEFAULT_BLOCK_PATH
131 block_umask = block_umask or DEFAULT_BLOCK_UMASK
132 #queue_module = queue_module or DEFAULT_QUEUE_MODULE
133 #queue_connection = queue_connection or DEFAULT_QUEUE_CONNECTION
135 self.hash_algorithm = 'sha256'
136 self.block_size = 4 * 1024 * 1024 # 4MB
138 self.default_policy = {'quota': DEFAULT_QUOTA, 'versioning': DEFAULT_VERSIONING}
142 return sys.modules[m]
144 self.db_module = load_module(db_module)
145 self.wrapper = self.db_module.DBWrapper(db_connection)
146 params = {'wrapper': self.wrapper}
147 self.permissions = self.db_module.Permissions(**params)
148 for x in ['READ', 'WRITE']:
149 setattr(self, x, getattr(self.db_module, x))
150 self.node = self.db_module.Node(**params)
151 for x in ['ROOTNODE', 'SERIAL', 'HASH', 'SIZE', 'TYPE', 'MTIME', 'MUSER', 'UUID', 'CHECKSUM', 'CLUSTER', 'MATCH_PREFIX', 'MATCH_EXACT']:
152 setattr(self, x, getattr(self.db_module, x))
154 self.block_module = load_module(block_module)
155 params = {'path': block_path,
156 'block_size': self.block_size,
157 'hash_algorithm': self.hash_algorithm,
158 'umask': block_umask}
159 self.store = self.block_module.Store(**params)
161 if queue_module and queue_connection:
162 self.queue_module = load_module(queue_module)
163 params = {'exchange': queue_connection,
164 'client_id': QUEUE_CLIENT_ID}
165 self.queue = self.queue_module.Queue(**params)
168 def send(self, *args):
174 self.queue = NoQueue()
181 def list_accounts(self, user, marker=None, limit=10000):
182 """Return a list of accounts the user can access."""
184 logger.debug("list_accounts: %s %s %s", user, marker, limit)
185 allowed = self._allowed_accounts(user)
186 start, limit = self._list_limits(allowed, marker, limit)
187 return allowed[start:start + limit]
190 def get_account_meta(self, user, account, domain, until=None, include_user_defined=True):
191 """Return a dictionary with the account metadata for the domain."""
193 logger.debug("get_account_meta: %s %s %s %s", user, account, domain, until)
194 path, node = self._lookup_account(account, user == account)
196 if until or node is None or account not in self._allowed_accounts(user):
197 raise NotAllowedError
199 props = self._get_properties(node, until)
200 mtime = props[self.MTIME]
204 count, bytes, tstamp = self._get_statistics(node, until)
205 tstamp = max(tstamp, mtime)
209 modified = self._get_statistics(node)[2] # Overall last modification.
210 modified = max(modified, mtime)
213 meta = {'name': account}
216 if props is not None and include_user_defined:
217 meta.update(dict(self.node.attribute_get(props[self.SERIAL], domain)))
218 if until is not None:
219 meta.update({'until_timestamp': tstamp})
220 meta.update({'name': account, 'count': count, 'bytes': bytes})
221 meta.update({'modified': modified})
225 def update_account_meta(self, user, account, domain, meta, replace=False):
226 """Update the metadata associated with the account for the domain."""
228 logger.debug("update_account_meta: %s %s %s %s %s", user, account, domain, meta, replace)
230 raise NotAllowedError
231 path, node = self._lookup_account(account, True)
232 self._put_metadata(user, node, domain, meta, replace)
235 def get_account_groups(self, user, account):
236 """Return a dictionary with the user groups defined for this account."""
238 logger.debug("get_account_groups: %s %s", user, account)
240 if account not in self._allowed_accounts(user):
241 raise NotAllowedError
243 self._lookup_account(account, True)
244 return self.permissions.group_dict(account)
247 def update_account_groups(self, user, account, groups, replace=False):
248 """Update the groups associated with the account."""
250 logger.debug("update_account_groups: %s %s %s %s", user, account, groups, replace)
252 raise NotAllowedError
253 self._lookup_account(account, True)
254 self._check_groups(groups)
256 self.permissions.group_destroy(account)
257 for k, v in groups.iteritems():
258 if not replace: # If not already deleted.
259 self.permissions.group_delete(account, k)
261 self.permissions.group_addmany(account, k, v)
264 def get_account_policy(self, user, account):
265 """Return a dictionary with the account policy."""
267 logger.debug("get_account_policy: %s %s", user, account)
269 if account not in self._allowed_accounts(user):
270 raise NotAllowedError
272 path, node = self._lookup_account(account, True)
273 return self._get_policy(node)
276 def update_account_policy(self, user, account, policy, replace=False):
277 """Update the policy associated with the account."""
279 logger.debug("update_account_policy: %s %s %s %s", user, account, policy, replace)
281 raise NotAllowedError
282 path, node = self._lookup_account(account, True)
283 self._check_policy(policy)
284 self._put_policy(node, policy, replace)
287 def put_account(self, user, account, policy={}):
288 """Create a new account with the given name."""
290 logger.debug("put_account: %s %s %s", user, account, policy)
292 raise NotAllowedError
293 node = self.node.node_lookup(account)
295 raise NameError('Account already exists')
297 self._check_policy(policy)
298 node = self._put_path(user, self.ROOTNODE, account)
299 self._put_policy(node, policy, True)
302 def delete_account(self, user, account):
303 """Delete the account with the given name."""
305 logger.debug("delete_account: %s %s", user, account)
307 raise NotAllowedError
308 node = self.node.node_lookup(account)
311 if not self.node.node_remove(node):
312 raise IndexError('Account is not empty')
313 self.permissions.group_destroy(account)
316 def list_containers(self, user, account, marker=None, limit=10000, shared=False, until=None, public=False):
317 """Return a list of containers existing under an account."""
319 logger.debug("list_containers: %s %s %s %s %s %s %s", user, account, marker, limit, shared, until, public)
321 if until or account not in self._allowed_accounts(user):
322 raise NotAllowedError
323 allowed = self._allowed_containers(user, account)
324 start, limit = self._list_limits(allowed, marker, limit)
325 return allowed[start:start + limit]
329 allowed.extend([x.split('/', 2)[1] for x in self.permissions.access_list_shared(account)])
331 allowed.extend([x[0].split('/', 2)[1] for x in self.permissions.public_list(account)])
332 allowed = list(set(allowed))
334 start, limit = self._list_limits(allowed, marker, limit)
335 return allowed[start:start + limit]
336 node = self.node.node_lookup(account)
337 containers = [x[0] for x in self._list_object_properties(node, account, '', '/', marker, limit, False, None, [], until)]
338 start, limit = self._list_limits([x[0] for x in containers], marker, limit)
339 return containers[start:start + limit]
342 def list_container_meta(self, user, account, container, domain, until=None):
343 """Return a list with all the container's object meta keys for the domain."""
345 logger.debug("list_container_meta: %s %s %s %s %s", user, account, container, domain, until)
349 raise NotAllowedError
350 allowed = self.permissions.access_list_paths(user, '/'.join((account, container)))
352 raise NotAllowedError
353 path, node = self._lookup_container(account, container)
354 before = until if until is not None else inf
355 allowed = self._get_formatted_paths(allowed)
356 return self.node.latest_attribute_keys(node, domain, before, CLUSTER_DELETED, allowed)
359 def get_container_meta(self, user, account, container, domain, until=None, include_user_defined=True):
360 """Return a dictionary with the container metadata for the domain."""
362 logger.debug("get_container_meta: %s %s %s %s %s", user, account, container, domain, until)
364 if until or container not in self._allowed_containers(user, account):
365 raise NotAllowedError
366 path, node = self._lookup_container(account, container)
367 props = self._get_properties(node, until)
368 mtime = props[self.MTIME]
369 count, bytes, tstamp = self._get_statistics(node, until)
370 tstamp = max(tstamp, mtime)
374 modified = self._get_statistics(node)[2] # Overall last modification.
375 modified = max(modified, mtime)
378 meta = {'name': container}
381 if include_user_defined:
382 meta.update(dict(self.node.attribute_get(props[self.SERIAL], domain)))
383 if until is not None:
384 meta.update({'until_timestamp': tstamp})
385 meta.update({'name': container, 'count': count, 'bytes': bytes})
386 meta.update({'modified': modified})
390 def update_container_meta(self, user, account, container, domain, meta, replace=False):
391 """Update the metadata associated with the container for the domain."""
393 logger.debug("update_container_meta: %s %s %s %s %s %s", user, account, container, domain, meta, replace)
395 raise NotAllowedError
396 path, node = self._lookup_container(account, container)
397 src_version_id, dest_version_id = self._put_metadata(user, node, domain, meta, replace)
398 if src_version_id is not None:
399 versioning = self._get_policy(node)['versioning']
400 if versioning != 'auto':
401 self.node.version_remove(src_version_id)
404 def get_container_policy(self, user, account, container):
405 """Return a dictionary with the container policy."""
407 logger.debug("get_container_policy: %s %s %s", user, account, container)
409 if container not in self._allowed_containers(user, account):
410 raise NotAllowedError
412 path, node = self._lookup_container(account, container)
413 return self._get_policy(node)
416 def update_container_policy(self, user, account, container, policy, replace=False):
417 """Update the policy associated with the container."""
419 logger.debug("update_container_policy: %s %s %s %s %s", user, account, container, policy, replace)
421 raise NotAllowedError
422 path, node = self._lookup_container(account, container)
423 self._check_policy(policy)
424 self._put_policy(node, policy, replace)
427 def put_container(self, user, account, container, policy={}):
428 """Create a new container with the given name."""
430 logger.debug("put_container: %s %s %s %s", user, account, container, policy)
432 raise NotAllowedError
434 path, node = self._lookup_container(account, container)
438 raise NameError('Container already exists')
440 self._check_policy(policy)
441 path = '/'.join((account, container))
442 node = self._put_path(user, self._lookup_account(account, True)[1], path)
443 self._put_policy(node, policy, True)
446 def delete_container(self, user, account, container, until=None, prefix='', delimiter=None):
447 """Delete/purge the container with the given name."""
449 logger.debug("delete_container: %s %s %s %s %s %s", user, account, container, until, prefix, delimiter)
451 raise NotAllowedError
452 path, node = self._lookup_container(account, container)
454 if until is not None:
455 hashes, size = self.node.node_purge_children(node, until, CLUSTER_HISTORY)
457 self.store.map_delete(h)
458 self.node.node_purge_children(node, until, CLUSTER_DELETED)
459 self._report_size_change(user, account, -size, {'action': 'container purge'})
462 if self._get_statistics(node)[0] > 0:
463 raise IndexError('Container is not empty')
464 hashes, size = self.node.node_purge_children(node, inf, CLUSTER_HISTORY)
466 self.store.map_delete(h)
467 self.node.node_purge_children(node, inf, CLUSTER_DELETED)
468 self.node.node_remove(node)
469 self._report_size_change(user, account, -size, {'action': 'container delete'})
471 def _list_objects(self, user, account, container, prefix, delimiter, marker, limit, virtual, domain, keys, shared, until, size_range, all_props, public):
472 if user != account and until:
473 raise NotAllowedError
474 if shared and public:
476 shared = self._list_object_permissions(user, account, container, prefix, shared=True, public=False)
479 path, node = self._lookup_container(account, container)
480 shared = self._get_formatted_paths(shared)
481 objects = self._list_object_properties(node, path, prefix, delimiter, marker, limit, virtual, domain, keys, until, size_range, shared, all_props)
484 objects.extend(self._list_public_object_properties(user, account, container, prefix, all_props))
486 objects.sort(key=lambda x: x[0])
487 start, limit = self._list_limits([x[0] for x in objects], marker, limit)
488 return objects[start:start + limit]
490 objects = self._list_public_object_properties(user, account, container, prefix, all_props)
491 start, limit = self._list_limits([x[0] for x in objects], marker, limit)
492 return objects[start:start + limit]
494 allowed = self._list_object_permissions(user, account, container, prefix, shared, public)
495 if shared and not allowed:
497 path, node = self._lookup_container(account, container)
498 allowed = self._get_formatted_paths(allowed)
499 objects = self._list_object_properties(node, path, prefix, delimiter, marker, limit, virtual, domain, keys, until, size_range, allowed, all_props)
500 start, limit = self._list_limits([x[0] for x in objects], marker, limit)
501 return objects[start:start + limit]
503 def _list_public_object_properties(self, user, account, container, prefix, all_props):
504 public = self._list_object_permissions(user, account, container, prefix, shared=False, public=True)
505 paths, nodes = self._lookup_objects(public)
506 path = '/'.join((account, container))
507 cont_prefix = path + '/'
508 paths = [x[len(cont_prefix):] for x in paths]
509 props = self.node.version_lookup_bulk(nodes, all_props=all_props)
510 objects = [(path,) + props for path, props in zip(paths, props)]
513 def _list_objects_no_limit(self, user, account, container, prefix, delimiter, virtual, domain, keys, shared, until, size_range, all_props, public):
516 marker = objects[-1] if objects else None
518 l = self._list_objects(user, account, container, prefix, delimiter, marker, limit, virtual, domain, keys, shared, until, size_range, all_props, public)
520 if not l or len(l) < limit:
524 def _list_object_permissions(self, user, account, container, prefix, shared, public):
526 path = '/'.join((account, container, prefix)).rstrip('/')
528 allowed = self.permissions.access_list_paths(user, path)
530 raise NotAllowedError
534 allowed.extend(self.permissions.access_list_shared(path))
536 allowed.extend([x[0] for x in self.permissions.public_list(path)])
537 allowed = list(set(allowed))
544 def list_objects(self, user, account, container, prefix='', delimiter=None, marker=None, limit=10000, virtual=True, domain=None, keys=[], shared=False, until=None, size_range=None, public=False):
545 """Return a list of object (name, version_id) tuples existing under a container."""
547 logger.debug("list_objects: %s %s %s %s %s %s %s %s %s %s %s %s %s %s", user, account, container, prefix, delimiter, marker, limit, virtual, domain, keys, shared, until, size_range, public)
548 return self._list_objects(user, account, container, prefix, delimiter, marker, limit, virtual, domain, keys, shared, until, size_range, False, public)
551 def list_object_meta(self, user, account, container, prefix='', delimiter=None, marker=None, limit=10000, virtual=True, domain=None, keys=[], shared=False, until=None, size_range=None, public=False):
552 """Return a list of object metadata dicts existing under a container."""
554 logger.debug("list_object_meta: %s %s %s %s %s %s %s %s %s %s %s %s %s %s", user, account, container, prefix, delimiter, marker, limit, virtual, domain, keys, shared, until, size_range, public)
555 props = self._list_objects(user, account, container, prefix, delimiter, marker, limit, virtual, domain, keys, shared, until, size_range, True, public)
559 objects.append({'subdir': p[0]})
561 objects.append({'name': p[0],
562 'bytes': p[self.SIZE + 1],
563 'type': p[self.TYPE + 1],
564 'hash': p[self.HASH + 1],
565 'version': p[self.SERIAL + 1],
566 'version_timestamp': p[self.MTIME + 1],
567 'modified': p[self.MTIME + 1] if until is None else None,
568 'modified_by': p[self.MUSER + 1],
569 'uuid': p[self.UUID + 1],
570 'checksum': p[self.CHECKSUM + 1]})
574 def list_object_permissions(self, user, account, container, prefix=''):
575 """Return a list of paths that enforce permissions under a container."""
577 logger.debug("list_object_permissions: %s %s %s %s", user, account, container, prefix)
578 return self._list_object_permissions(user, account, container, prefix, True, False)
581 def list_object_public(self, user, account, container, prefix=''):
582 """Return a dict mapping paths to public ids for objects that are public under a container."""
584 logger.debug("list_object_public: %s %s %s %s", user, account, container, prefix)
586 for path, p in self.permissions.public_list('/'.join((account, container, prefix))):
587 public[path] = p + ULTIMATE_ANSWER
591 def get_object_meta(self, user, account, container, name, domain, version=None, include_user_defined=True):
592 """Return a dictionary with the object metadata for the domain."""
594 logger.debug("get_object_meta: %s %s %s %s %s %s", user, account, container, name, domain, version)
595 self._can_read(user, account, container, name)
596 path, node = self._lookup_object(account, container, name)
597 props = self._get_version(node, version)
599 modified = props[self.MTIME]
602 modified = self._get_version(node)[self.MTIME] # Overall last modification.
603 except NameError: # Object may be deleted.
604 del_props = self.node.version_lookup(node, inf, CLUSTER_DELETED)
605 if del_props is None:
606 raise NameError('Object does not exist')
607 modified = del_props[self.MTIME]
610 if include_user_defined:
611 meta.update(dict(self.node.attribute_get(props[self.SERIAL], domain)))
612 meta.update({'name': name,
613 'bytes': props[self.SIZE],
614 'type': props[self.TYPE],
615 'hash': props[self.HASH],
616 'version': props[self.SERIAL],
617 'version_timestamp': props[self.MTIME],
618 'modified': modified,
619 'modified_by': props[self.MUSER],
620 'uuid': props[self.UUID],
621 'checksum': props[self.CHECKSUM]})
625 def update_object_meta(self, user, account, container, name, domain, meta, replace=False):
626 """Update the metadata associated with the object for the domain and return the new version."""
628 logger.debug("update_object_meta: %s %s %s %s %s %s %s", user, account, container, name, domain, meta, replace)
629 self._can_write(user, account, container, name)
630 path, node = self._lookup_object(account, container, name)
631 src_version_id, dest_version_id = self._put_metadata(user, node, domain, meta, replace)
632 self._apply_versioning(account, container, src_version_id)
633 return dest_version_id
636 def get_object_permissions(self, user, account, container, name):
637 """Return the action allowed on the object, the path
638 from which the object gets its permissions from,
639 along with a dictionary containing the permissions."""
641 logger.debug("get_object_permissions: %s %s %s %s", user, account, container, name)
643 permissions_path = self._get_permissions_path(account, container, name)
645 if self.permissions.access_check(permissions_path, self.WRITE, user):
647 elif self.permissions.access_check(permissions_path, self.READ, user):
650 raise NotAllowedError
651 self._lookup_object(account, container, name)
652 return (allowed, permissions_path, self.permissions.access_get(permissions_path))
655 def update_object_permissions(self, user, account, container, name, permissions):
656 """Update the permissions associated with the object."""
658 logger.debug("update_object_permissions: %s %s %s %s %s", user, account, container, name, permissions)
660 raise NotAllowedError
661 path = self._lookup_object(account, container, name)[0]
662 self._check_permissions(path, permissions)
663 self.permissions.access_set(path, permissions)
664 self._report_sharing_change(user, account, path, {'members':self.permissions.access_members(path)})
667 def get_object_public(self, user, account, container, name):
668 """Return the public id of the object if applicable."""
670 logger.debug("get_object_public: %s %s %s %s", user, account, container, name)
671 self._can_read(user, account, container, name)
672 path = self._lookup_object(account, container, name)[0]
673 p = self.permissions.public_get(path)
679 def update_object_public(self, user, account, container, name, public):
680 """Update the public status of the object."""
682 logger.debug("update_object_public: %s %s %s %s %s", user, account, container, name, public)
683 self._can_write(user, account, container, name)
684 path = self._lookup_object(account, container, name)[0]
686 self.permissions.public_unset(path)
688 self.permissions.public_set(path)
691 def get_object_hashmap(self, user, account, container, name, version=None):
692 """Return the object's size and a list with partial hashes."""
694 logger.debug("get_object_hashmap: %s %s %s %s %s", user, account, container, name, version)
695 self._can_read(user, account, container, name)
696 path, node = self._lookup_object(account, container, name)
697 props = self._get_version(node, version)
698 hashmap = self.store.map_get(binascii.unhexlify(props[self.HASH]))
699 return props[self.SIZE], [binascii.hexlify(x) for x in hashmap]
701 def _update_object_hash(self, user, account, container, name, size, type, hash, checksum, domain, meta, replace_meta, permissions, src_node=None, src_version_id=None, is_copy=False):
702 if permissions is not None and user != account:
703 raise NotAllowedError
704 self._can_write(user, account, container, name)
705 if permissions is not None:
706 path = '/'.join((account, container, name))
707 self._check_permissions(path, permissions)
709 account_path, account_node = self._lookup_account(account, True)
710 container_path, container_node = self._lookup_container(account, container)
711 path, node = self._put_object_node(container_path, container_node, name)
712 pre_version_id, dest_version_id = self._put_version_duplicate(user, node, src_node=src_node, size=size, type=type, hash=hash, checksum=checksum, is_copy=is_copy)
715 if src_version_id is None:
716 src_version_id = pre_version_id
717 self._put_metadata_duplicate(src_version_id, dest_version_id, domain, meta, replace_meta)
720 del_size = self._apply_versioning(account, container, pre_version_id)
721 size_delta = size - del_size
723 account_quota = long(self._get_policy(account_node)['quota'])
724 container_quota = long(self._get_policy(container_node)['quota'])
725 if (account_quota > 0 and self._get_statistics(account_node)[1] + size_delta > account_quota) or \
726 (container_quota > 0 and self._get_statistics(container_node)[1] + size_delta > container_quota):
727 # This must be executed in a transaction, so the version is never created if it fails.
729 self._report_size_change(user, account, size_delta, {'action': 'object update'})
731 if permissions is not None:
732 self.permissions.access_set(path, permissions)
733 self._report_sharing_change(user, account, path, {'members':self.permissions.access_members(path)})
735 self._report_object_change(user, account, path, details={'version': dest_version_id, 'action': 'object update'})
736 return dest_version_id
739 def update_object_hashmap(self, user, account, container, name, size, type, hashmap, checksum, domain, meta={}, replace_meta=False, permissions=None):
740 """Create/update an object with the specified size and partial hashes."""
742 logger.debug("update_object_hashmap: %s %s %s %s %s %s %s %s", user, account, container, name, size, type, hashmap, checksum)
743 if size == 0: # No such thing as an empty hashmap.
744 hashmap = [self.put_block('')]
745 map = HashMap(self.block_size, self.hash_algorithm)
746 map.extend([binascii.unhexlify(x) for x in hashmap])
747 missing = self.store.block_search(map)
750 ie.data = [binascii.hexlify(x) for x in missing]
754 dest_version_id = self._update_object_hash(user, account, container, name, size, type, binascii.hexlify(hash), checksum, domain, meta, replace_meta, permissions)
755 self.store.map_put(hash, map)
756 return dest_version_id
759 def update_object_checksum(self, user, account, container, name, version, checksum):
760 """Update an object's checksum."""
762 logger.debug("update_object_checksum: %s %s %s %s %s %s", user, account, container, name, version, checksum)
763 # Update objects with greater version and same hashmap and size (fix metadata updates).
764 self._can_write(user, account, container, name)
765 path, node = self._lookup_object(account, container, name)
766 props = self._get_version(node, version)
767 versions = self.node.node_get_versions(node)
769 if x[self.SERIAL] >= int(version) and x[self.HASH] == props[self.HASH] and x[self.SIZE] == props[self.SIZE]:
770 self.node.version_put_property(x[self.SERIAL], 'checksum', checksum)
772 def _copy_object(self, user, src_account, src_container, src_name, dest_account, dest_container, dest_name, type, dest_domain=None, dest_meta={}, replace_meta=False, permissions=None, src_version=None, is_move=False, delimiter=None):
773 dest_version_ids = []
774 self._can_read(user, src_account, src_container, src_name)
775 path, node = self._lookup_object(src_account, src_container, src_name)
776 # TODO: Will do another fetch of the properties in duplicate version...
777 props = self._get_version(node, src_version) # Check to see if source exists.
778 src_version_id = props[self.SERIAL]
779 hash = props[self.HASH]
780 size = props[self.SIZE]
781 is_copy = not is_move and (src_account, src_container, src_name) != (dest_account, dest_container, dest_name) # New uuid.
782 dest_version_ids.append(self._update_object_hash(user, dest_account, dest_container, dest_name, size, type, hash, None, dest_domain, dest_meta, replace_meta, permissions, src_node=node, src_version_id=src_version_id, is_copy=is_copy))
784 self._delete_object(user, src_account, src_container, src_name)
787 prefix = src_name + delimiter if not src_name.endswith(delimiter) else src_name
788 src_names = self._list_objects_no_limit(user, src_account, src_container, prefix, delimiter=None, virtual=True, domain=None, keys=[], shared=False, until=None, size_range=None, all_props=True, public=False)
789 paths = [elem[0] for elem in src_names]
790 nodes = [elem[2] for elem in src_names]
791 # TODO: Will do another fetch of the properties in duplicate version...
792 props = self._get_versions(nodes) # Check to see if source exists.
794 for prop, path, node in zip(props, paths, nodes):
795 src_version_id = prop[self.SERIAL]
796 hash = prop[self.HASH]
797 vtype = prop[self.TYPE]
798 dest_prefix = dest_name + delimiter if not dest_name.endswith(delimiter) else dest_name
799 vdest_name = path.replace(prefix, dest_prefix, 1)
800 dest_version_ids.append(self._update_object_hash(user, dest_account, dest_container, vdest_name, size, vtype, hash, None, dest_domain, dest_meta, replace_meta, permissions, src_node=node, src_version_id=src_version_id, is_copy=is_copy))
802 self._delete_object(user, src_account, src_container, path)
803 return dest_version_ids[0] if len(dest_version_ids) == 1 else dest_version_ids
806 def copy_object(self, user, src_account, src_container, src_name, dest_account, dest_container, dest_name, type, domain, meta={}, replace_meta=False, permissions=None, src_version=None, delimiter=None):
807 """Copy an object's data and metadata."""
809 logger.debug("copy_object: %s %s %s %s %s %s %s %s %s %s %s %s %s %s", user, src_account, src_container, src_name, dest_account, dest_container, dest_name, type, domain, meta, replace_meta, permissions, src_version, delimiter)
810 dest_version_id = self._copy_object(user, src_account, src_container, src_name, dest_account, dest_container, dest_name, type, domain, meta, replace_meta, permissions, src_version, False, delimiter)
811 return dest_version_id
814 def move_object(self, user, src_account, src_container, src_name, dest_account, dest_container, dest_name, type, domain, meta={}, replace_meta=False, permissions=None, delimiter=None):
815 """Move an object's data and metadata."""
817 logger.debug("move_object: %s %s %s %s %s %s %s %s %s %s %s %s %s", user, src_account, src_container, src_name, dest_account, dest_container, dest_name, type, domain, meta, replace_meta, permissions, delimiter)
818 if user != src_account:
819 raise NotAllowedError
820 dest_version_id = self._copy_object(user, src_account, src_container, src_name, dest_account, dest_container, dest_name, type, domain, meta, replace_meta, permissions, None, True, delimiter)
821 return dest_version_id
823 def _delete_object(self, user, account, container, name, until=None, delimiter=None):
825 raise NotAllowedError
827 if until is not None:
828 path = '/'.join((account, container, name))
829 node = self.node.node_lookup(path)
834 h, s = self.node.node_purge(node, until, CLUSTER_NORMAL)
837 h, s = self.node.node_purge(node, until, CLUSTER_HISTORY)
841 self.store.map_delete(h)
842 self.node.node_purge(node, until, CLUSTER_DELETED)
844 props = self._get_version(node)
846 self.permissions.access_clear(path)
847 self._report_size_change(user, account, -size, {'action': 'object purge'})
850 path, node = self._lookup_object(account, container, name)
851 src_version_id, dest_version_id = self._put_version_duplicate(user, node, size=0, type='', hash=None, checksum='', cluster=CLUSTER_DELETED)
852 del_size = self._apply_versioning(account, container, src_version_id)
854 self._report_size_change(user, account, -del_size, {'action': 'object delete'})
855 self._report_object_change(user, account, path, details={'action': 'object delete'})
856 self.permissions.access_clear(path)
859 prefix = name + delimiter if not name.endswith(delimiter) else name
860 src_names = self._list_objects_no_limit(user, account, container, prefix, delimiter=None, virtual=True, domain=None, keys=[], shared=False, until=None, size_range=None, all_props=True, public=False)
863 path = '/'.join((account, container, t[0]))
865 src_version_id, dest_version_id = self._put_version_duplicate(user, node, size=0, type='', hash=None, checksum='', cluster=CLUSTER_DELETED)
866 del_size = self._apply_versioning(account, container, src_version_id)
868 self._report_size_change(user, account, -del_size, {'action': 'object delete'})
869 self._report_object_change(user, account, path, details={'action': 'object delete'})
871 self.permissions.access_clear_bulk(paths)
874 def delete_object(self, user, account, container, name, until=None, prefix='', delimiter=None):
875 """Delete/purge an object."""
877 logger.debug("delete_object: %s %s %s %s %s %s %s", user, account, container, name, until, prefix, delimiter)
878 self._delete_object(user, account, container, name, until, delimiter)
881 def list_versions(self, user, account, container, name):
882 """Return a list of all (version, version_timestamp) tuples for an object."""
884 logger.debug("list_versions: %s %s %s %s", user, account, container, name)
885 self._can_read(user, account, container, name)
886 path, node = self._lookup_object(account, container, name)
887 versions = self.node.node_get_versions(node)
888 return [[x[self.SERIAL], x[self.MTIME]] for x in versions if x[self.CLUSTER] != CLUSTER_DELETED]
891 def get_uuid(self, user, uuid):
892 """Return the (account, container, name) for the UUID given."""
894 logger.debug("get_uuid: %s %s", user, uuid)
895 info = self.node.latest_uuid(uuid)
899 account, container, name = path.split('/', 2)
900 self._can_read(user, account, container, name)
901 return (account, container, name)
904 def get_public(self, user, public):
905 """Return the (account, container, name) for the public id given."""
907 logger.debug("get_public: %s %s", user, public)
908 if public is None or public < ULTIMATE_ANSWER:
910 path = self.permissions.public_path(public - ULTIMATE_ANSWER)
913 account, container, name = path.split('/', 2)
914 self._can_read(user, account, container, name)
915 return (account, container, name)
917 @backend_method(autocommit=0)
918 def get_block(self, hash):
919 """Return a block's data."""
921 logger.debug("get_block: %s", hash)
922 block = self.store.block_get(binascii.unhexlify(hash))
924 raise NameError('Block does not exist')
927 @backend_method(autocommit=0)
928 def put_block(self, data):
929 """Store a block and return the hash."""
931 logger.debug("put_block: %s", len(data))
932 return binascii.hexlify(self.store.block_put(data))
934 @backend_method(autocommit=0)
935 def update_block(self, hash, data, offset=0):
936 """Update a known block and return the hash."""
938 logger.debug("update_block: %s %s %s", hash, len(data), offset)
939 if offset == 0 and len(data) == self.block_size:
940 return self.put_block(data)
941 h = self.store.block_update(binascii.unhexlify(hash), offset, data)
942 return binascii.hexlify(h)
946 def _generate_uuid(self):
947 return str(uuidlib.uuid4())
949 def _put_object_node(self, path, parent, name):
950 path = '/'.join((path, name))
951 node = self.node.node_lookup(path)
953 node = self.node.node_create(parent, path)
956 def _put_path(self, user, parent, path):
957 node = self.node.node_create(parent, path)
958 self.node.version_create(node, None, 0, '', None, user, self._generate_uuid(), '', CLUSTER_NORMAL)
961 def _lookup_account(self, account, create=True):
962 node = self.node.node_lookup(account)
963 if node is None and create:
964 node = self._put_path(account, self.ROOTNODE, account) # User is account.
967 def _lookup_container(self, account, container):
968 path = '/'.join((account, container))
969 node = self.node.node_lookup(path)
971 raise NameError('Container does not exist')
974 def _lookup_object(self, account, container, name):
975 path = '/'.join((account, container, name))
976 node = self.node.node_lookup(path)
978 raise NameError('Object does not exist')
981 def _lookup_objects(self, paths):
982 nodes = self.node.node_lookup_bulk(paths)
984 raise NameError('Object does not exist')
987 def _get_properties(self, node, until=None):
988 """Return properties until the timestamp given."""
990 before = until if until is not None else inf
991 props = self.node.version_lookup(node, before, CLUSTER_NORMAL)
992 if props is None and until is not None:
993 props = self.node.version_lookup(node, before, CLUSTER_HISTORY)
995 raise NameError('Path does not exist')
998 def _get_statistics(self, node, until=None):
999 """Return count, sum of size and latest timestamp of everything under node."""
1002 stats = self.node.statistics_get(node, CLUSTER_NORMAL)
1004 stats = self.node.statistics_latest(node, until, CLUSTER_DELETED)
1009 def _get_version(self, node, version=None):
1011 props = self.node.version_lookup(node, inf, CLUSTER_NORMAL)
1013 raise NameError('Object does not exist')
1016 version = int(version)
1018 raise IndexError('Version does not exist')
1019 props = self.node.version_get_properties(version)
1020 if props is None or props[self.CLUSTER] == CLUSTER_DELETED:
1021 raise IndexError('Version does not exist')
1024 def _get_versions(self, nodes, version=None):
1026 props = self.node.version_lookup_bulk(nodes, inf, CLUSTER_NORMAL)
1028 raise NameError('Object does not exist')
1031 version = int(version)
1033 raise IndexError('Version does not exist')
1034 props = self.node.version_get_properties(version)
1035 if props is None or props[self.CLUSTER] == CLUSTER_DELETED:
1036 raise IndexError('Version does not exist')
1039 def _put_version_duplicate(self, user, node, src_node=None, size=None, type=None, hash=None, checksum=None, cluster=CLUSTER_NORMAL, is_copy=False):
1040 """Create a new version of the node."""
1042 props = self.node.version_lookup(node if src_node is None else src_node, inf, CLUSTER_NORMAL)
1043 if props is not None:
1044 src_version_id = props[self.SERIAL]
1045 src_hash = props[self.HASH]
1046 src_size = props[self.SIZE]
1047 src_type = props[self.TYPE]
1048 src_checksum = props[self.CHECKSUM]
1050 src_version_id = None
1055 if size is None: # Set metadata.
1056 hash = src_hash # This way hash can be set to None (account or container).
1060 if checksum is None:
1061 checksum = src_checksum
1062 uuid = self._generate_uuid() if (is_copy or src_version_id is None) else props[self.UUID]
1064 if src_node is None:
1065 pre_version_id = src_version_id
1067 pre_version_id = None
1068 props = self.node.version_lookup(node, inf, CLUSTER_NORMAL)
1069 if props is not None:
1070 pre_version_id = props[self.SERIAL]
1071 if pre_version_id is not None:
1072 self.node.version_recluster(pre_version_id, CLUSTER_HISTORY)
1074 dest_version_id, mtime = self.node.version_create(node, hash, size, type, src_version_id, user, uuid, checksum, cluster)
1075 return pre_version_id, dest_version_id
1077 def _put_metadata_duplicate(self, src_version_id, dest_version_id, domain, meta, replace=False):
1078 if src_version_id is not None:
1079 self.node.attribute_copy(src_version_id, dest_version_id)
1081 self.node.attribute_del(dest_version_id, domain, (k for k, v in meta.iteritems() if v == ''))
1082 self.node.attribute_set(dest_version_id, domain, ((k, v) for k, v in meta.iteritems() if v != ''))
1084 self.node.attribute_del(dest_version_id, domain)
1085 self.node.attribute_set(dest_version_id, domain, ((k, v) for k, v in meta.iteritems()))
1087 def _put_metadata(self, user, node, domain, meta, replace=False):
1088 """Create a new version and store metadata."""
1090 src_version_id, dest_version_id = self._put_version_duplicate(user, node)
1091 self._put_metadata_duplicate(src_version_id, dest_version_id, domain, meta, replace)
1092 return src_version_id, dest_version_id
1094 def _list_limits(self, listing, marker, limit):
1098 start = listing.index(marker) + 1
1101 if not limit or limit > 10000:
1105 def _list_object_properties(self, parent, path, prefix='', delimiter=None, marker=None, limit=10000, virtual=True, domain=None, keys=[], until=None, size_range=None, allowed=[], all_props=False):
1106 cont_prefix = path + '/'
1107 prefix = cont_prefix + prefix
1108 start = cont_prefix + marker if marker else None
1109 before = until if until is not None else inf
1110 filterq = keys if domain else []
1113 objects, prefixes = self.node.latest_version_list(parent, prefix, delimiter, start, limit, before, CLUSTER_DELETED, allowed, domain, filterq, sizeq, all_props)
1114 objects.extend([(p, None) for p in prefixes] if virtual else [])
1115 objects.sort(key=lambda x: x[0])
1116 objects = [(x[0][len(cont_prefix):],) + x[1:] for x in objects]
1119 # Reporting functions.
1121 def _report_size_change(self, user, account, size, details={}):
1122 logger.debug("_report_size_change: %s %s %s %s", user, account, size, details)
1123 account_node = self._lookup_account(account, True)[1]
1124 total = self._get_statistics(account_node)[1]
1125 details.update({'user': user, 'total': total})
1126 self.messages.append((QUEUE_MESSAGE_KEY_PREFIX % ('resource.diskspace',), account, QUEUE_INSTANCE_ID, 'diskspace', float(size), details))
1128 def _report_object_change(self, user, account, path, details={}):
1129 logger.debug("_report_object_change: %s %s %s %s", user, account, path, details)
1130 details.update({'user': user})
1131 self.messages.append((QUEUE_MESSAGE_KEY_PREFIX % ('object',), account, QUEUE_INSTANCE_ID, 'object', path, details))
1133 def _report_sharing_change(self, user, account, path, details={}):
1134 logger.debug("_report_permissions_change: %s %s %s %s", user, account, path, details)
1135 details.update({'user': user})
1136 self.messages.append((QUEUE_MESSAGE_KEY_PREFIX % ('sharing',), account, QUEUE_INSTANCE_ID, 'sharing', path, details))
1140 def _check_policy(self, policy):
1141 for k in policy.keys():
1143 policy[k] = self.default_policy.get(k)
1144 for k, v in policy.iteritems():
1146 q = int(v) # May raise ValueError.
1149 elif k == 'versioning':
1150 if v not in ['auto', 'none']:
1155 def _put_policy(self, node, policy, replace):
1157 for k, v in self.default_policy.iteritems():
1160 self.node.policy_set(node, policy)
1162 def _get_policy(self, node):
1163 policy = self.default_policy.copy()
1164 policy.update(self.node.policy_get(node))
1167 def _apply_versioning(self, account, container, version_id):
1168 """Delete the provided version if such is the policy.
1169 Return size of object removed.
1172 if version_id is None:
1174 path, node = self._lookup_container(account, container)
1175 versioning = self._get_policy(node)['versioning']
1176 if versioning != 'auto':
1177 hash, size = self.node.version_remove(version_id)
1178 self.store.map_delete(hash)
1182 # Access control functions.
1184 def _check_groups(self, groups):
1185 # raise ValueError('Bad characters in groups')
1188 def _check_permissions(self, path, permissions):
1189 # raise ValueError('Bad characters in permissions')
1192 def _get_formatted_paths(self, paths):
1195 node = self.node.node_lookup(p)
1196 if node is not None:
1197 props = self.node.version_lookup(node, inf, CLUSTER_NORMAL)
1198 if props is not None:
1199 if props[self.TYPE].split(';', 1)[0].strip() in ('application/directory', 'application/folder'):
1200 formatted.append((p.rstrip('/') + '/', self.MATCH_PREFIX))
1201 formatted.append((p, self.MATCH_EXACT))
1204 def _get_permissions_path(self, account, container, name):
1205 path = '/'.join((account, container, name))
1206 permission_paths = self.permissions.access_inherit(path)
1207 permission_paths.sort()
1208 permission_paths.reverse()
1209 for p in permission_paths:
1213 if p.count('/') < 2:
1215 node = self.node.node_lookup(p)
1216 if node is not None:
1217 props = self.node.version_lookup(node, inf, CLUSTER_NORMAL)
1218 if props is not None:
1219 if props[self.TYPE].split(';', 1)[0].strip() in ('application/directory', 'application/folder'):
1223 def _can_read(self, user, account, container, name):
1226 path = '/'.join((account, container, name))
1227 if self.permissions.public_get(path) is not None:
1229 path = self._get_permissions_path(account, container, name)
1231 raise NotAllowedError
1232 if not self.permissions.access_check(path, self.READ, user) and not self.permissions.access_check(path, self.WRITE, user):
1233 raise NotAllowedError
1235 def _can_write(self, user, account, container, name):
1238 path = '/'.join((account, container, name))
1239 path = self._get_permissions_path(account, container, name)
1241 raise NotAllowedError
1242 if not self.permissions.access_check(path, self.WRITE, user):
1243 raise NotAllowedError
1245 def _allowed_accounts(self, user):
1247 for path in self.permissions.access_list_paths(user):
1248 allow.add(path.split('/', 1)[0])
1249 return sorted(allow)
1251 def _allowed_containers(self, user, account):
1253 for path in self.permissions.access_list_paths(user, account):
1254 allow.add(path.split('/', 2)[1])
1255 return sorted(allow)