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, \
43 AccountExists, ContainerExists, AccountNotEmpty, ContainerNotEmpty, ItemNotExists, VersionNotExists
45 # Stripped-down version of the HashMap class found in tools.
48 def __init__(self, blocksize, blockhash):
49 super(HashMap, self).__init__()
50 self.blocksize = blocksize
51 self.blockhash = blockhash
53 def _hash_raw(self, v):
54 h = hashlib.new(self.blockhash)
60 return self._hash_raw('')
62 return self.__getitem__(0)
68 h += [('\x00' * len(h[0]))] * (s - len(h))
70 h = [self._hash_raw(h[x] + h[x + 1]) for x in range(0, len(h), 2)]
73 # Default modules and settings.
74 DEFAULT_DB_MODULE = 'pithos.backends.lib.sqlalchemy'
75 DEFAULT_DB_CONNECTION = 'sqlite:///backend.db'
76 DEFAULT_BLOCK_MODULE = 'pithos.backends.lib.hashfiler'
77 DEFAULT_BLOCK_PATH = 'data/'
78 DEFAULT_BLOCK_UMASK = 0o022
79 #DEFAULT_QUEUE_MODULE = 'pithos.backends.lib.rabbitmq'
80 #DEFAULT_QUEUE_CONNECTION = 'rabbitmq://guest:guest@localhost:5672/pithos'
82 QUEUE_MESSAGE_KEY_PREFIX = 'pithos.%s'
83 QUEUE_CLIENT_ID = 'pithos'
84 QUEUE_INSTANCE_ID = '1'
86 ( CLUSTER_NORMAL, CLUSTER_HISTORY, CLUSTER_DELETED ) = range(3)
93 logger = logging.getLogger(__name__)
96 def backend_method(func=None, autocommit=1):
99 return backend_method(func, autocommit)
104 def fn(self, *args, **kw):
105 self.wrapper.execute()
108 ret = func(self, *args, **kw)
109 for m in self.messages:
111 self.wrapper.commit()
114 self.wrapper.rollback()
119 class ModularBackend(BaseBackend):
120 """A modular backend.
122 Uses modules for SQL functions and storage.
125 def __init__(self, db_module=None, db_connection=None,
126 block_module=None, block_path=None, block_umask=None,
127 queue_module=None, queue_connection=None):
128 db_module = db_module or DEFAULT_DB_MODULE
129 db_connection = db_connection or DEFAULT_DB_CONNECTION
130 block_module = block_module or DEFAULT_BLOCK_MODULE
131 block_path = block_path or DEFAULT_BLOCK_PATH
132 block_umask = block_umask or DEFAULT_BLOCK_UMASK
133 #queue_module = queue_module or DEFAULT_QUEUE_MODULE
134 #queue_connection = queue_connection or DEFAULT_QUEUE_CONNECTION
136 self.hash_algorithm = 'sha256'
137 self.block_size = 4 * 1024 * 1024 # 4MB
139 self.default_policy = {'quota': DEFAULT_QUOTA, 'versioning': DEFAULT_VERSIONING}
143 return sys.modules[m]
145 self.db_module = load_module(db_module)
146 self.wrapper = self.db_module.DBWrapper(db_connection)
147 params = {'wrapper': self.wrapper}
148 self.permissions = self.db_module.Permissions(**params)
149 for x in ['READ', 'WRITE']:
150 setattr(self, x, getattr(self.db_module, x))
151 self.node = self.db_module.Node(**params)
152 for x in ['ROOTNODE', 'SERIAL', 'HASH', 'SIZE', 'TYPE', 'MTIME', 'MUSER', 'UUID', 'CHECKSUM', 'CLUSTER', 'MATCH_PREFIX', 'MATCH_EXACT']:
153 setattr(self, x, getattr(self.db_module, x))
155 self.block_module = load_module(block_module)
156 params = {'path': block_path,
157 'block_size': self.block_size,
158 'hash_algorithm': self.hash_algorithm,
159 'umask': block_umask}
160 self.store = self.block_module.Store(**params)
162 if queue_module and queue_connection:
163 self.queue_module = load_module(queue_module)
164 params = {'exchange': queue_connection,
165 'client_id': QUEUE_CLIENT_ID}
166 self.queue = self.queue_module.Queue(**params)
169 def send(self, *args):
175 self.queue = NoQueue()
182 def list_accounts(self, user, marker=None, limit=10000):
183 """Return a list of accounts the user can access."""
185 logger.debug("list_accounts: %s %s %s", user, marker, limit)
186 allowed = self._allowed_accounts(user)
187 start, limit = self._list_limits(allowed, marker, limit)
188 return allowed[start:start + limit]
191 def get_account_meta(self, user, account, domain, until=None, include_user_defined=True):
192 """Return a dictionary with the account metadata for the domain."""
194 logger.debug("get_account_meta: %s %s %s %s", user, account, domain, until)
195 path, node = self._lookup_account(account, user == account)
197 if until or node is None or account not in self._allowed_accounts(user):
198 raise NotAllowedError
200 props = self._get_properties(node, until)
201 mtime = props[self.MTIME]
205 count, bytes, tstamp = self._get_statistics(node, until)
206 tstamp = max(tstamp, mtime)
210 modified = self._get_statistics(node)[2] # Overall last modification.
211 modified = max(modified, mtime)
214 meta = {'name': account}
217 if props is not None and include_user_defined:
218 meta.update(dict(self.node.attribute_get(props[self.SERIAL], domain)))
219 if until is not None:
220 meta.update({'until_timestamp': tstamp})
221 meta.update({'name': account, 'count': count, 'bytes': bytes})
222 meta.update({'modified': modified})
226 def update_account_meta(self, user, account, domain, meta, replace=False):
227 """Update the metadata associated with the account for the domain."""
229 logger.debug("update_account_meta: %s %s %s %s %s", user, account, domain, meta, replace)
231 raise NotAllowedError
232 path, node = self._lookup_account(account, True)
233 self._put_metadata(user, node, domain, meta, replace)
236 def get_account_groups(self, user, account):
237 """Return a dictionary with the user groups defined for this account."""
239 logger.debug("get_account_groups: %s %s", user, account)
241 if account not in self._allowed_accounts(user):
242 raise NotAllowedError
244 self._lookup_account(account, True)
245 return self.permissions.group_dict(account)
248 def update_account_groups(self, user, account, groups, replace=False):
249 """Update the groups associated with the account."""
251 logger.debug("update_account_groups: %s %s %s %s", user, account, groups, replace)
253 raise NotAllowedError
254 self._lookup_account(account, True)
255 self._check_groups(groups)
257 self.permissions.group_destroy(account)
258 for k, v in groups.iteritems():
259 if not replace: # If not already deleted.
260 self.permissions.group_delete(account, k)
262 self.permissions.group_addmany(account, k, v)
265 def get_account_policy(self, user, account):
266 """Return a dictionary with the account policy."""
268 logger.debug("get_account_policy: %s %s", user, account)
270 if account not in self._allowed_accounts(user):
271 raise NotAllowedError
273 path, node = self._lookup_account(account, True)
274 return self._get_policy(node)
277 def update_account_policy(self, user, account, policy, replace=False):
278 """Update the policy associated with the account."""
280 logger.debug("update_account_policy: %s %s %s %s", user, account, policy, replace)
282 raise NotAllowedError
283 path, node = self._lookup_account(account, True)
284 self._check_policy(policy)
285 self._put_policy(node, policy, replace)
288 def put_account(self, user, account, policy={}):
289 """Create a new account with the given name."""
291 logger.debug("put_account: %s %s %s", user, account, policy)
293 raise NotAllowedError
294 node = self.node.node_lookup(account)
296 raise AccountExists('Account already exists')
298 self._check_policy(policy)
299 node = self._put_path(user, self.ROOTNODE, account)
300 self._put_policy(node, policy, True)
303 def delete_account(self, user, account):
304 """Delete the account with the given name."""
306 logger.debug("delete_account: %s %s", user, account)
308 raise NotAllowedError
309 node = self.node.node_lookup(account)
312 if not self.node.node_remove(node):
313 raise AccountNotEmpty('Account is not empty')
314 self.permissions.group_destroy(account)
317 def list_containers(self, user, account, marker=None, limit=10000, shared=False, until=None, public=False):
318 """Return a list of containers existing under an account."""
320 logger.debug("list_containers: %s %s %s %s %s %s %s", user, account, marker, limit, shared, until, public)
322 if until or account not in self._allowed_accounts(user):
323 raise NotAllowedError
324 allowed = self._allowed_containers(user, account)
325 start, limit = self._list_limits(allowed, marker, limit)
326 return allowed[start:start + limit]
330 allowed.update([x.split('/', 2)[1] for x in self.permissions.access_list_shared(account)])
332 allowed.update([x[0].split('/', 2)[1] for x in self.permissions.public_list(account)])
333 allowed = sorted(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 ContainerExists('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'})
463 if self._get_statistics(node)[0] > 0:
464 raise ContainerNotEmpty('Container is not empty')
465 hashes, size = self.node.node_purge_children(node, inf, CLUSTER_HISTORY)
467 self.store.map_delete(h)
468 self.node.node_purge_children(node, inf, CLUSTER_DELETED)
469 self.node.node_remove(node)
470 self._report_size_change(user, account, -size, {'action': 'container delete'})
472 # remove only contents
473 src_names = self._list_objects_no_limit(user, account, container, prefix='', delimiter=None, virtual=False, domain=None, keys=[], shared=False, until=None, size_range=None, all_props=True, public=False)
476 path = '/'.join((account, container, t[0]))
478 src_version_id, dest_version_id = self._put_version_duplicate(user, node, size=0, type='', hash=None, checksum='', cluster=CLUSTER_DELETED)
479 del_size = self._apply_versioning(account, container, src_version_id)
481 self._report_size_change(user, account, -del_size, {'action': 'object delete'})
482 self._report_object_change(user, account, path, details={'action': 'object delete'})
484 self.permissions.access_clear_bulk(paths)
486 def _list_objects(self, user, account, container, prefix, delimiter, marker, limit, virtual, domain, keys, shared, until, size_range, all_props, public):
487 if user != account and until:
488 raise NotAllowedError
489 if shared and public:
491 shared = self._list_object_permissions(user, account, container, prefix, shared=True, public=False)
494 path, node = self._lookup_container(account, container)
495 shared = self._get_formatted_paths(shared)
496 objects = self._list_object_properties(node, path, prefix, delimiter, marker, limit, virtual, domain, keys, until, size_range, shared, all_props)
499 objects.extend(self._list_public_object_properties(user, account, container, prefix, all_props))
501 objects.sort(key=lambda x: x[0])
502 start, limit = self._list_limits([x[0] for x in objects], marker, limit)
503 return objects[start:start + limit]
505 objects = self._list_public_object_properties(user, account, container, prefix, all_props)
506 start, limit = self._list_limits([x[0] for x in objects], marker, limit)
507 return objects[start:start + limit]
509 allowed = self._list_object_permissions(user, account, container, prefix, shared, public)
510 if shared and not allowed:
512 path, node = self._lookup_container(account, container)
513 allowed = self._get_formatted_paths(allowed)
514 objects = self._list_object_properties(node, path, prefix, delimiter, marker, limit, virtual, domain, keys, until, size_range, allowed, all_props)
515 start, limit = self._list_limits([x[0] for x in objects], marker, limit)
516 return objects[start:start + limit]
518 def _list_public_object_properties(self, user, account, container, prefix, all_props):
519 public = self._list_object_permissions(user, account, container, prefix, shared=False, public=True)
520 paths, nodes = self._lookup_objects(public)
521 path = '/'.join((account, container))
522 cont_prefix = path + '/'
523 paths = [x[len(cont_prefix):] for x in paths]
524 props = self.node.version_lookup_bulk(nodes, all_props=all_props)
525 objects = [(path,) + props for path, props in zip(paths, props)]
528 def _list_objects_no_limit(self, user, account, container, prefix, delimiter, virtual, domain, keys, shared, until, size_range, all_props, public):
531 marker = objects[-1] if objects else None
533 l = self._list_objects(user, account, container, prefix, delimiter, marker, limit, virtual, domain, keys, shared, until, size_range, all_props, public)
535 if not l or len(l) < limit:
539 def _list_object_permissions(self, user, account, container, prefix, shared, public):
541 path = '/'.join((account, container, prefix)).rstrip('/')
543 allowed = self.permissions.access_list_paths(user, path)
545 raise NotAllowedError
549 allowed.update(self.permissions.access_list_shared(path))
551 allowed.update([x[0] for x in self.permissions.public_list(path)])
552 allowed = sorted(allowed)
558 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):
559 """Return a list of object (name, version_id) tuples existing under a container."""
561 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)
562 return self._list_objects(user, account, container, prefix, delimiter, marker, limit, virtual, domain, keys, shared, until, size_range, False, public)
565 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):
566 """Return a list of object metadata dicts existing under a container."""
568 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)
569 props = self._list_objects(user, account, container, prefix, delimiter, marker, limit, virtual, domain, keys, shared, until, size_range, True, public)
573 objects.append({'subdir': p[0]})
575 objects.append({'name': p[0],
576 'bytes': p[self.SIZE + 1],
577 'type': p[self.TYPE + 1],
578 'hash': p[self.HASH + 1],
579 'version': p[self.SERIAL + 1],
580 'version_timestamp': p[self.MTIME + 1],
581 'modified': p[self.MTIME + 1] if until is None else None,
582 'modified_by': p[self.MUSER + 1],
583 'uuid': p[self.UUID + 1],
584 'checksum': p[self.CHECKSUM + 1]})
588 def list_object_permissions(self, user, account, container, prefix=''):
589 """Return a list of paths that enforce permissions under a container."""
591 logger.debug("list_object_permissions: %s %s %s %s", user, account, container, prefix)
592 return self._list_object_permissions(user, account, container, prefix, True, False)
595 def list_object_public(self, user, account, container, prefix=''):
596 """Return a dict mapping paths to public ids for objects that are public under a container."""
598 logger.debug("list_object_public: %s %s %s %s", user, account, container, prefix)
600 for path, p in self.permissions.public_list('/'.join((account, container, prefix))):
601 public[path] = p + ULTIMATE_ANSWER
605 def get_object_meta(self, user, account, container, name, domain, version=None, include_user_defined=True):
606 """Return a dictionary with the object metadata for the domain."""
608 logger.debug("get_object_meta: %s %s %s %s %s %s", user, account, container, name, domain, version)
609 self._can_read(user, account, container, name)
610 path, node = self._lookup_object(account, container, name)
611 props = self._get_version(node, version)
613 modified = props[self.MTIME]
616 modified = self._get_version(node)[self.MTIME] # Overall last modification.
617 except NameError: # Object may be deleted.
618 del_props = self.node.version_lookup(node, inf, CLUSTER_DELETED)
619 if del_props is None:
620 raise ItemNotExists('Object does not exist')
621 modified = del_props[self.MTIME]
624 if include_user_defined:
625 meta.update(dict(self.node.attribute_get(props[self.SERIAL], domain)))
626 meta.update({'name': name,
627 'bytes': props[self.SIZE],
628 'type': props[self.TYPE],
629 'hash': props[self.HASH],
630 'version': props[self.SERIAL],
631 'version_timestamp': props[self.MTIME],
632 'modified': modified,
633 'modified_by': props[self.MUSER],
634 'uuid': props[self.UUID],
635 'checksum': props[self.CHECKSUM]})
639 def update_object_meta(self, user, account, container, name, domain, meta, replace=False):
640 """Update the metadata associated with the object for the domain and return the new version."""
642 logger.debug("update_object_meta: %s %s %s %s %s %s %s", user, account, container, name, domain, meta, replace)
643 self._can_write(user, account, container, name)
644 path, node = self._lookup_object(account, container, name)
645 src_version_id, dest_version_id = self._put_metadata(user, node, domain, meta, replace)
646 self._apply_versioning(account, container, src_version_id)
647 return dest_version_id
650 def get_object_permissions(self, user, account, container, name):
651 """Return the action allowed on the object, the path
652 from which the object gets its permissions from,
653 along with a dictionary containing the permissions."""
655 logger.debug("get_object_permissions: %s %s %s %s", user, account, container, name)
657 permissions_path = self._get_permissions_path(account, container, name)
659 if self.permissions.access_check(permissions_path, self.WRITE, user):
661 elif self.permissions.access_check(permissions_path, self.READ, user):
664 raise NotAllowedError
665 self._lookup_object(account, container, name)
666 return (allowed, permissions_path, self.permissions.access_get(permissions_path))
669 def update_object_permissions(self, user, account, container, name, permissions):
670 """Update the permissions associated with the object."""
672 logger.debug("update_object_permissions: %s %s %s %s %s", user, account, container, name, permissions)
674 raise NotAllowedError
675 path = self._lookup_object(account, container, name)[0]
676 self._check_permissions(path, permissions)
677 self.permissions.access_set(path, permissions)
678 self._report_sharing_change(user, account, path, {'members':self.permissions.access_members(path)})
681 def get_object_public(self, user, account, container, name):
682 """Return the public id of the object if applicable."""
684 logger.debug("get_object_public: %s %s %s %s", user, account, container, name)
685 self._can_read(user, account, container, name)
686 path = self._lookup_object(account, container, name)[0]
687 p = self.permissions.public_get(path)
693 def update_object_public(self, user, account, container, name, public):
694 """Update the public status of the object."""
696 logger.debug("update_object_public: %s %s %s %s %s", user, account, container, name, public)
697 self._can_write(user, account, container, name)
698 path = self._lookup_object(account, container, name)[0]
700 self.permissions.public_unset(path)
702 self.permissions.public_set(path)
705 def get_object_hashmap(self, user, account, container, name, version=None):
706 """Return the object's size and a list with partial hashes."""
708 logger.debug("get_object_hashmap: %s %s %s %s %s", user, account, container, name, version)
709 self._can_read(user, account, container, name)
710 path, node = self._lookup_object(account, container, name)
711 props = self._get_version(node, version)
712 hashmap = self.store.map_get(binascii.unhexlify(props[self.HASH]))
713 return props[self.SIZE], [binascii.hexlify(x) for x in hashmap]
715 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):
716 if permissions is not None and user != account:
717 raise NotAllowedError
718 self._can_write(user, account, container, name)
719 if permissions is not None:
720 path = '/'.join((account, container, name))
721 self._check_permissions(path, permissions)
723 account_path, account_node = self._lookup_account(account, True)
724 container_path, container_node = self._lookup_container(account, container)
725 path, node = self._put_object_node(container_path, container_node, name)
726 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)
729 if src_version_id is None:
730 src_version_id = pre_version_id
731 self._put_metadata_duplicate(src_version_id, dest_version_id, domain, meta, replace_meta)
734 del_size = self._apply_versioning(account, container, pre_version_id)
735 size_delta = size - del_size
737 account_quota = long(self._get_policy(account_node)['quota'])
738 container_quota = long(self._get_policy(container_node)['quota'])
739 if (account_quota > 0 and self._get_statistics(account_node)[1] + size_delta > account_quota) or \
740 (container_quota > 0 and self._get_statistics(container_node)[1] + size_delta > container_quota):
741 # This must be executed in a transaction, so the version is never created if it fails.
743 self._report_size_change(user, account, size_delta, {'action': 'object update'})
745 if permissions is not None:
746 self.permissions.access_set(path, permissions)
747 self._report_sharing_change(user, account, path, {'members':self.permissions.access_members(path)})
749 self._report_object_change(user, account, path, details={'version': dest_version_id, 'action': 'object update'})
750 return dest_version_id
753 def update_object_hashmap(self, user, account, container, name, size, type, hashmap, checksum, domain, meta={}, replace_meta=False, permissions=None):
754 """Create/update an object with the specified size and partial hashes."""
756 logger.debug("update_object_hashmap: %s %s %s %s %s %s %s %s", user, account, container, name, size, type, hashmap, checksum)
757 if size == 0: # No such thing as an empty hashmap.
758 hashmap = [self.put_block('')]
759 map = HashMap(self.block_size, self.hash_algorithm)
760 map.extend([binascii.unhexlify(x) for x in hashmap])
761 missing = self.store.block_search(map)
764 ie.data = [binascii.hexlify(x) for x in missing]
768 dest_version_id = self._update_object_hash(user, account, container, name, size, type, binascii.hexlify(hash), checksum, domain, meta, replace_meta, permissions)
769 self.store.map_put(hash, map)
770 return dest_version_id
773 def update_object_checksum(self, user, account, container, name, version, checksum):
774 """Update an object's checksum."""
776 logger.debug("update_object_checksum: %s %s %s %s %s %s", user, account, container, name, version, checksum)
777 # Update objects with greater version and same hashmap and size (fix metadata updates).
778 self._can_write(user, account, container, name)
779 path, node = self._lookup_object(account, container, name)
780 props = self._get_version(node, version)
781 versions = self.node.node_get_versions(node)
783 if x[self.SERIAL] >= int(version) and x[self.HASH] == props[self.HASH] and x[self.SIZE] == props[self.SIZE]:
784 self.node.version_put_property(x[self.SERIAL], 'checksum', checksum)
786 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):
787 dest_version_ids = []
788 self._can_read(user, src_account, src_container, src_name)
789 path, node = self._lookup_object(src_account, src_container, src_name)
790 # TODO: Will do another fetch of the properties in duplicate version...
791 props = self._get_version(node, src_version) # Check to see if source exists.
792 src_version_id = props[self.SERIAL]
793 hash = props[self.HASH]
794 size = props[self.SIZE]
795 is_copy = not is_move and (src_account, src_container, src_name) != (dest_account, dest_container, dest_name) # New uuid.
796 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))
797 if is_move and (src_account, src_container, src_name) != (dest_account, dest_container, dest_name):
798 self._delete_object(user, src_account, src_container, src_name)
801 prefix = src_name + delimiter if not src_name.endswith(delimiter) else src_name
802 src_names = self._list_objects_no_limit(user, src_account, src_container, prefix, delimiter=None, virtual=False, domain=None, keys=[], shared=False, until=None, size_range=None, all_props=True, public=False)
803 src_names.sort(key=lambda x: x[2]) # order by nodes
804 paths = [elem[0] for elem in src_names]
805 nodes = [elem[2] for elem in src_names]
806 # TODO: Will do another fetch of the properties in duplicate version...
807 props = self._get_versions(nodes) # Check to see if source exists.
809 for prop, path, node in zip(props, paths, nodes):
810 src_version_id = prop[self.SERIAL]
811 hash = prop[self.HASH]
812 vtype = prop[self.TYPE]
813 size = prop[self.SIZE]
814 dest_prefix = dest_name + delimiter if not dest_name.endswith(delimiter) else dest_name
815 vdest_name = path.replace(prefix, dest_prefix, 1)
816 dest_version_ids.append(self._update_object_hash(user, dest_account, dest_container, vdest_name, size, vtype, hash, None, dest_domain, meta={}, replace_meta=False, permissions=None, src_node=node, src_version_id=src_version_id, is_copy=is_copy))
817 if is_move and (src_account, src_container, src_name) != (dest_account, dest_container, dest_name):
818 self._delete_object(user, src_account, src_container, path)
819 return dest_version_ids[0] if len(dest_version_ids) == 1 else dest_version_ids
822 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):
823 """Copy an object's data and metadata."""
825 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)
826 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)
827 return dest_version_id
830 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):
831 """Move an object's data and metadata."""
833 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)
834 if user != src_account:
835 raise NotAllowedError
836 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)
837 return dest_version_id
839 def _delete_object(self, user, account, container, name, until=None, delimiter=None):
841 raise NotAllowedError
843 if until is not None:
844 path = '/'.join((account, container, name))
845 node = self.node.node_lookup(path)
850 h, s = self.node.node_purge(node, until, CLUSTER_NORMAL)
853 h, s = self.node.node_purge(node, until, CLUSTER_HISTORY)
857 self.store.map_delete(h)
858 self.node.node_purge(node, until, CLUSTER_DELETED)
860 props = self._get_version(node)
862 self.permissions.access_clear(path)
863 self._report_size_change(user, account, -size, {'action': 'object purge'})
866 path, node = self._lookup_object(account, container, name)
867 src_version_id, dest_version_id = self._put_version_duplicate(user, node, size=0, type='', hash=None, checksum='', cluster=CLUSTER_DELETED)
868 del_size = self._apply_versioning(account, container, src_version_id)
870 self._report_size_change(user, account, -del_size, {'action': 'object delete'})
871 self._report_object_change(user, account, path, details={'action': 'object delete'})
872 self.permissions.access_clear(path)
875 prefix = name + delimiter if not name.endswith(delimiter) else name
876 src_names = self._list_objects_no_limit(user, account, container, prefix, delimiter=None, virtual=False, domain=None, keys=[], shared=False, until=None, size_range=None, all_props=True, public=False)
879 path = '/'.join((account, container, t[0]))
881 src_version_id, dest_version_id = self._put_version_duplicate(user, node, size=0, type='', hash=None, checksum='', cluster=CLUSTER_DELETED)
882 del_size = self._apply_versioning(account, container, src_version_id)
884 self._report_size_change(user, account, -del_size, {'action': 'object delete'})
885 self._report_object_change(user, account, path, details={'action': 'object delete'})
887 self.permissions.access_clear_bulk(paths)
890 def delete_object(self, user, account, container, name, until=None, prefix='', delimiter=None):
891 """Delete/purge an object."""
893 logger.debug("delete_object: %s %s %s %s %s %s %s", user, account, container, name, until, prefix, delimiter)
894 self._delete_object(user, account, container, name, until, delimiter)
897 def list_versions(self, user, account, container, name):
898 """Return a list of all (version, version_timestamp) tuples for an object."""
900 logger.debug("list_versions: %s %s %s %s", user, account, container, name)
901 self._can_read(user, account, container, name)
902 path, node = self._lookup_object(account, container, name)
903 versions = self.node.node_get_versions(node)
904 return [[x[self.SERIAL], x[self.MTIME]] for x in versions if x[self.CLUSTER] != CLUSTER_DELETED]
907 def get_uuid(self, user, uuid):
908 """Return the (account, container, name) for the UUID given."""
910 logger.debug("get_uuid: %s %s", user, uuid)
911 info = self.node.latest_uuid(uuid)
915 account, container, name = path.split('/', 2)
916 self._can_read(user, account, container, name)
917 return (account, container, name)
920 def get_public(self, user, public):
921 """Return the (account, container, name) for the public id given."""
923 logger.debug("get_public: %s %s", user, public)
924 if public is None or public < ULTIMATE_ANSWER:
926 path = self.permissions.public_path(public - ULTIMATE_ANSWER)
929 account, container, name = path.split('/', 2)
930 self._can_read(user, account, container, name)
931 return (account, container, name)
933 @backend_method(autocommit=0)
934 def get_block(self, hash):
935 """Return a block's data."""
937 logger.debug("get_block: %s", hash)
938 block = self.store.block_get(binascii.unhexlify(hash))
940 raise ItemNotExists('Block does not exist')
943 @backend_method(autocommit=0)
944 def put_block(self, data):
945 """Store a block and return the hash."""
947 logger.debug("put_block: %s", len(data))
948 return binascii.hexlify(self.store.block_put(data))
950 @backend_method(autocommit=0)
951 def update_block(self, hash, data, offset=0):
952 """Update a known block and return the hash."""
954 logger.debug("update_block: %s %s %s", hash, len(data), offset)
955 if offset == 0 and len(data) == self.block_size:
956 return self.put_block(data)
957 h = self.store.block_update(binascii.unhexlify(hash), offset, data)
958 return binascii.hexlify(h)
962 def _generate_uuid(self):
963 return str(uuidlib.uuid4())
965 def _put_object_node(self, path, parent, name):
966 path = '/'.join((path, name))
967 node = self.node.node_lookup(path)
969 node = self.node.node_create(parent, path)
972 def _put_path(self, user, parent, path):
973 node = self.node.node_create(parent, path)
974 self.node.version_create(node, None, 0, '', None, user, self._generate_uuid(), '', CLUSTER_NORMAL)
977 def _lookup_account(self, account, create=True):
978 node = self.node.node_lookup(account)
979 if node is None and create:
980 node = self._put_path(account, self.ROOTNODE, account) # User is account.
983 def _lookup_container(self, account, container):
984 path = '/'.join((account, container))
985 node = self.node.node_lookup(path)
987 raise ItemNotExists('Container does not exist')
990 def _lookup_object(self, account, container, name):
991 path = '/'.join((account, container, name))
992 node = self.node.node_lookup(path)
994 raise ItemNotExists('Object does not exist')
997 def _lookup_objects(self, paths):
998 nodes = self.node.node_lookup_bulk(paths)
1001 def _get_properties(self, node, until=None):
1002 """Return properties until the timestamp given."""
1004 before = until if until is not None else inf
1005 props = self.node.version_lookup(node, before, CLUSTER_NORMAL)
1006 if props is None and until is not None:
1007 props = self.node.version_lookup(node, before, CLUSTER_HISTORY)
1009 raise ItemNotExists('Path does not exist')
1012 def _get_statistics(self, node, until=None):
1013 """Return count, sum of size and latest timestamp of everything under node."""
1016 stats = self.node.statistics_get(node, CLUSTER_NORMAL)
1018 stats = self.node.statistics_latest(node, until, CLUSTER_DELETED)
1023 def _get_version(self, node, version=None):
1025 props = self.node.version_lookup(node, inf, CLUSTER_NORMAL)
1027 raise ItemNotExists('Object does not exist')
1030 version = int(version)
1032 raise VersionNotExists('Version does not exist')
1033 props = self.node.version_get_properties(version)
1034 if props is None or props[self.CLUSTER] == CLUSTER_DELETED:
1035 raise VersionNotExists('Version does not exist')
1038 def _get_versions(self, nodes):
1039 return self.node.version_lookup_bulk(nodes, inf, CLUSTER_NORMAL)
1041 def _put_version_duplicate(self, user, node, src_node=None, size=None, type=None, hash=None, checksum=None, cluster=CLUSTER_NORMAL, is_copy=False):
1042 """Create a new version of the node."""
1044 props = self.node.version_lookup(node if src_node is None else src_node, inf, CLUSTER_NORMAL)
1045 if props is not None:
1046 src_version_id = props[self.SERIAL]
1047 src_hash = props[self.HASH]
1048 src_size = props[self.SIZE]
1049 src_type = props[self.TYPE]
1050 src_checksum = props[self.CHECKSUM]
1052 src_version_id = None
1057 if size is None: # Set metadata.
1058 hash = src_hash # This way hash can be set to None (account or container).
1062 if checksum is None:
1063 checksum = src_checksum
1064 uuid = self._generate_uuid() if (is_copy or src_version_id is None) else props[self.UUID]
1066 if src_node is None:
1067 pre_version_id = src_version_id
1069 pre_version_id = None
1070 props = self.node.version_lookup(node, inf, CLUSTER_NORMAL)
1071 if props is not None:
1072 pre_version_id = props[self.SERIAL]
1073 if pre_version_id is not None:
1074 self.node.version_recluster(pre_version_id, CLUSTER_HISTORY)
1076 dest_version_id, mtime = self.node.version_create(node, hash, size, type, src_version_id, user, uuid, checksum, cluster)
1077 return pre_version_id, dest_version_id
1079 def _put_metadata_duplicate(self, src_version_id, dest_version_id, domain, meta, replace=False):
1080 if src_version_id is not None:
1081 self.node.attribute_copy(src_version_id, dest_version_id)
1083 self.node.attribute_del(dest_version_id, domain, (k for k, v in meta.iteritems() if v == ''))
1084 self.node.attribute_set(dest_version_id, domain, ((k, v) for k, v in meta.iteritems() if v != ''))
1086 self.node.attribute_del(dest_version_id, domain)
1087 self.node.attribute_set(dest_version_id, domain, ((k, v) for k, v in meta.iteritems()))
1089 def _put_metadata(self, user, node, domain, meta, replace=False):
1090 """Create a new version and store metadata."""
1092 src_version_id, dest_version_id = self._put_version_duplicate(user, node)
1093 self._put_metadata_duplicate(src_version_id, dest_version_id, domain, meta, replace)
1094 return src_version_id, dest_version_id
1096 def _list_limits(self, listing, marker, limit):
1100 start = listing.index(marker) + 1
1103 if not limit or limit > 10000:
1107 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):
1108 cont_prefix = path + '/'
1109 prefix = cont_prefix + prefix
1110 start = cont_prefix + marker if marker else None
1111 before = until if until is not None else inf
1112 filterq = keys if domain else []
1115 objects, prefixes = self.node.latest_version_list(parent, prefix, delimiter, start, limit, before, CLUSTER_DELETED, allowed, domain, filterq, sizeq, all_props)
1116 objects.extend([(p, None) for p in prefixes] if virtual else [])
1117 objects.sort(key=lambda x: x[0])
1118 objects = [(x[0][len(cont_prefix):],) + x[1:] for x in objects]
1121 # Reporting functions.
1123 def _report_size_change(self, user, account, size, details={}):
1124 logger.debug("_report_size_change: %s %s %s %s", user, account, size, details)
1125 account_node = self._lookup_account(account, True)[1]
1126 total = self._get_statistics(account_node)[1]
1127 details.update({'user': user, 'total': total})
1128 self.messages.append((QUEUE_MESSAGE_KEY_PREFIX % ('resource.diskspace',), account, QUEUE_INSTANCE_ID, 'diskspace', float(size), details))
1130 def _report_object_change(self, user, account, path, details={}):
1131 logger.debug("_report_object_change: %s %s %s %s", user, account, path, details)
1132 details.update({'user': user})
1133 self.messages.append((QUEUE_MESSAGE_KEY_PREFIX % ('object',), account, QUEUE_INSTANCE_ID, 'object', path, details))
1135 def _report_sharing_change(self, user, account, path, details={}):
1136 logger.debug("_report_permissions_change: %s %s %s %s", user, account, path, details)
1137 details.update({'user': user})
1138 self.messages.append((QUEUE_MESSAGE_KEY_PREFIX % ('sharing',), account, QUEUE_INSTANCE_ID, 'sharing', path, details))
1142 def _check_policy(self, policy):
1143 for k in policy.keys():
1145 policy[k] = self.default_policy.get(k)
1146 for k, v in policy.iteritems():
1148 q = int(v) # May raise ValueError.
1151 elif k == 'versioning':
1152 if v not in ['auto', 'none']:
1157 def _put_policy(self, node, policy, replace):
1159 for k, v in self.default_policy.iteritems():
1162 self.node.policy_set(node, policy)
1164 def _get_policy(self, node):
1165 policy = self.default_policy.copy()
1166 policy.update(self.node.policy_get(node))
1169 def _apply_versioning(self, account, container, version_id):
1170 """Delete the provided version if such is the policy.
1171 Return size of object removed.
1174 if version_id is None:
1176 path, node = self._lookup_container(account, container)
1177 versioning = self._get_policy(node)['versioning']
1178 if versioning != 'auto':
1179 hash, size = self.node.version_remove(version_id)
1180 self.store.map_delete(hash)
1184 # Access control functions.
1186 def _check_groups(self, groups):
1187 # raise ValueError('Bad characters in groups')
1190 def _check_permissions(self, path, permissions):
1191 # raise ValueError('Bad characters in permissions')
1194 def _get_formatted_paths(self, paths):
1197 node = self.node.node_lookup(p)
1198 if node is not None:
1199 props = self.node.version_lookup(node, inf, CLUSTER_NORMAL)
1200 if props is not None:
1201 if props[self.TYPE].split(';', 1)[0].strip() in ('application/directory', 'application/folder'):
1202 formatted.append((p.rstrip('/') + '/', self.MATCH_PREFIX))
1203 formatted.append((p, self.MATCH_EXACT))
1206 def _get_permissions_path(self, account, container, name):
1207 path = '/'.join((account, container, name))
1208 permission_paths = self.permissions.access_inherit(path)
1209 permission_paths.sort()
1210 permission_paths.reverse()
1211 for p in permission_paths:
1215 if p.count('/') < 2:
1217 node = self.node.node_lookup(p)
1218 if node is not None:
1219 props = self.node.version_lookup(node, inf, CLUSTER_NORMAL)
1220 if props is not None:
1221 if props[self.TYPE].split(';', 1)[0].strip() in ('application/directory', 'application/folder'):
1225 def _can_read(self, user, account, container, name):
1228 path = '/'.join((account, container, name))
1229 if self.permissions.public_get(path) is not None:
1231 path = self._get_permissions_path(account, container, name)
1233 raise NotAllowedError
1234 if not self.permissions.access_check(path, self.READ, user) and not self.permissions.access_check(path, self.WRITE, user):
1235 raise NotAllowedError
1237 def _can_write(self, user, account, container, name):
1240 path = '/'.join((account, container, name))
1241 path = self._get_permissions_path(account, container, name)
1243 raise NotAllowedError
1244 if not self.permissions.access_check(path, self.WRITE, user):
1245 raise NotAllowedError
1247 def _allowed_accounts(self, user):
1249 for path in self.permissions.access_list_paths(user):
1250 allow.add(path.split('/', 1)[0])
1251 return sorted(allow)
1253 def _allowed_containers(self, user, account):
1255 for path in self.permissions.access_list_paths(user, account):
1256 allow.add(path.split('/', 2)[1])
1257 return sorted(allow)